A multi-seat vehicle door projection parallel emotional healing method based on emotional perception

Abstract

The application discloses a multi-seat vehicle door projection parallel emotional healing method based on emotion perception, relates to the technical field of intelligent vehicle cabin, and mainly comprises the following steps: obtaining the valence and arousal degree of each passenger through a cabin perception system; based on the valence, the saturation is nonlinearly modulated according to the Stevens power law, and based on the arousal degree, the total adjustment time length is calculated according to the Yerkes-Dodson law; the comprehensive priority is calculated by combining the seat safety priority and the emotional intensity obtained by weighting the valence deviation degree and the arousal degree; when the ratio of the highest priority to the second highest priority reaches a threshold value, the high-priority seat remains the original parameters, and the rest are degraded, otherwise, the dynamic speed of each seat is staggered, and the upper limit of the saturation is limited; and the dynamic light and shadow pattern is projected by controlling each vehicle door projection module. Through multi-seat independent perception and double-mode conflict resolution, the application realizes parallel and non-interfering personalized healing, and improves the accuracy of emotional regulation and the visual harmony in the cabin based on a nonlinear psychophysics model.

Description

Technical Field

[0001] This invention relates to the field of automotive intelligent cockpit technology, specifically to a multi-seat door projection parallel emotion healing method based on emotion perception. Background Technology

[0002] In the field of intelligent automotive cockpits, adaptive environmental adjustment based on occupant emotional states has become an important technological direction for improving the driving experience. Existing technologies include some solutions that collect occupant facial expressions, voice features, or physiological signals through cameras, radar, microphones, or biometric sensors within the cockpit. These sensors then utilize emotion recognition models to output emotion categories or dimensions (such as valence and arousal), thereby driving actuators in the vehicle, such as ambient lighting, audio systems, air conditioning, fragrance systems, and seat massagers, to generate multi-sensory stimulation to improve or match the occupant's emotional state. For example, some systems can automatically adjust the color and brightness of interior lighting, play soothing music, or suggest changes to navigation routes based on identified sadness or anger. Regarding emotion representation, some studies employ three-dimensional emotion space models that include valence, arousal, and dominance to more precisely describe emotional states. Furthermore, a few technological solutions are beginning to focus on the differentiated needs of multi-occupant scenarios, providing independent control of lighting or audio in different seating areas through cockpit partitioning, or compromising the emotional needs of multiple occupants into a single comprehensive adjustment command through weighted averaging or priority selection, thereby adjusting the cockpit environment as a whole or in specific areas. However, the aforementioned existing technologies still have significant limitations: On the one hand, most adjustment methods are limited to relatively simple visual feedback such as color gradation, brightness breathing, or abstract lighting effects, lacking concrete, dynamic light and shadow themes that can guide breathing rhythms and have been validated by psychological and biological affinity design. Furthermore, the mapping from emotion to light and shadow is mostly linear or follows fixed rules, failing to fully consider the nonlinear psychophysical characteristics of human perception of color saturation and intervention duration. On the other hand, in multi-occupant scenarios, existing solutions either adopt a globally uniform adjustment strategy for all seats, failing to provide truly parallel and non-interfering personalized healing for different occupants at the same time; or, while achieving independent control of zones, they ignore visual interference problems such as light crosstalk and flickering when multiple door projections or light and shadow devices work simultaneously, and lack a systematic conflict resolution mechanism. Therefore, how to achieve nonlinear light and shadow modulation based on psychological models and collaborative interference avoidance between multiple light and shadow paths while taking into account the differentiated emotional needs of multiple occupants has become a pressing technical problem that current in-vehicle emotional healing systems need to solve. Summary of the Invention

[0003] To simultaneously address the diverse emotional needs of multiple passengers and achieve nonlinear light and shadow modulation based on psychological models, as well as collaborative interference avoidance among multiple light and shadow paths, this invention proposes a parallel emotional therapy method based on emotion perception through multi-seat door projection, comprising the following steps: S1. The emotional state of each seat occupant during the sliding window time is obtained through the cockpit perception system. The emotional state includes at least the average valence representing the degree of emotional pleasure and the average arousal representing the level of physiological activation. S2. Calculate the dynamic light and shadow parameters corresponding to each seat according to the emotional state of each seat occupant. The dynamic light and shadow parameters include at least the saturation based on the average valence and nonlinearly modulated according to Stevens' power law, the total adjustment time calculated piecewise according to Yerkes-Dodson's law based on the average arousal, and the dynamic speed corresponding to the emotional state. S3. Calculate the overall priority of each seat based on the safety priority of each seat and the emotional intensity obtained by weighting the normalized deviation calculated from the average valence and the average arousal. S4. Adjust the dynamic lighting parameters according to the ratio of the highest priority to the second highest priority in the overall priority. When the ratio is greater than or equal to the preset threshold, keep the seat corresponding to the highest priority running according to the original dynamic lighting parameters and control the other seats to execute the downgrade mode; otherwise, adjust the dynamic speed in the dynamic lighting parameters of each seat to be staggered and limit the saturation to within the preset upper limit. S5. Based on the adjusted dynamic light and shadow parameters, control the projection modules installed on the inside of each door to project dynamic light and shadow patterns onto the corresponding occupants.

[0004] This invention introduces nonlinear modulation based on Stevens' power law and Yerkes-Dodson's law, as well as a dual-mode conflict resolution strategy based on priority and emotional intensity, to provide each seat with dynamic light and shadow therapy that matches the emotional state of the passengers in a multi-occupant parallel scenario. At the same time, it effectively reduces visual interference between different projections and improves the accuracy and overall comfort of personalized therapy.

[0005] Furthermore, in step S1, the cockpit perception system includes a wide-angle cockpit monitoring camera. Through spatial three-dimensional calibration and spatial weight filtering algorithms, the detected three-dimensional coordinates of the face are compared with the preset seat space domain. If the three-dimensional coordinates of the face fall within a certain seat space domain, the facial data of the face is marked as valid input for that seat and used for emotion recognition of that seat. If the three-dimensional coordinates of the face do not fall within any seat space domain or fall within an adjacent seat space domain, the facial data of the face is marked as invalid and discarded.

[0006] Furthermore, in step S2, the saturation based on the average valence nonlinearly modulated according to Stevens' power law is expressed by the following formula: In the formula, The average valence detected within the sliding window. The saturation level corresponding to the average valence. As the baseline saturation, For maximum saturation, Valence deviation, Based on the calm baseline valence, For the maximum valence deviation, Let be the power exponent of Stevens' power law, and when The saturation is fixed at time. .

[0007] Furthermore, in step S2, the total adjustment time, calculated piecewise based on the average arousal level according to the Yerkes-Dodson Law, is expressed by the following formula: In the formula, The average wakefulness detected within the sliding window. The total settling time corresponding to the average arousal level. For optimal wakefulness, These represent the minimum, medium, and maximum adjustment durations, respectively. The power exponent for low wake-up region. It is a power exponent of the high wake-up region, and .

[0008] Furthermore, in step S3, the overall priority of each seat is calculated, expressed by the formula: In the formula, The overall priority of the i-th seat. The preset safety priority weight for the i-th seat Let i be the normalized valence deviation of the i-th seat. Let i be the average valence of the i-th seat. Let i be the average arousal level of the i-th seat. These are the weighting coefficients, and .

[0009] Furthermore, following step S6, the following step is also included: The average valence and average arousal of the crew were re-collected, and the continuous reward value was calculated based on the changes before and after adjustment. The minimum adjustment duration was updated using an ε-greedy strategy and Q-learning. Medium adjustment duration Maximum adjustment duration Low wake-up zone power index High Arousal Zone Power Index Optimal wakefulness One or more of them.

[0010] Furthermore, in step S4, the downgrade mode specifically refers to: Using the baseline saturation as the minimum adjustment limit for saturation, the saturation of the remaining seats is adjusted to 50% of the saturation corresponding to the average valence, and the brightness of the remaining seats is reduced to 50% of the preset fixed brightness value corresponding to the mood category of that seat.

[0011] Furthermore, in step S4, the dynamic speeds of each seat are adjusted to be staggered from one another, specifically as follows: Detect the ratio of the dynamic cycles of any two seats. If the ratio falls within a preset range, multiply the dynamic cycle of one of the seats by a preset multiplier factor.

[0012] Furthermore, in step S6, a light-shielding grille is installed at the light outlet of the projection module. The light-shielding grille is used to limit the horizontal divergence angle of the dynamic light and shadow pattern within a preset angle range.

[0013] Furthermore, the door is provided with a dynamic light and shadow pattern projection area and coated with a directional diffuse reflection layer, which reflects the dynamic light and shadow pattern to the lateral field of vision of the corresponding occupant through directional diffuse reflection.

[0014] Compared with the prior art, the present invention has at least the following beneficial effects: (1) The present invention proposes a parallel emotional healing method based on emotion perception of multiple seats and door projection. By equipping each seat with an independent door projection module and combining it with a spatial weight filtering algorithm, it realizes the non-crosstalk parallel acquisition and independent healing of the emotional state of multiple passengers under a single camera, so that passengers in different seats at the same time can obtain personalized light and shadow intervention without interference. (2) Based on Stevens' power law, nonlinear saturation modulation and based on Yerkes-Dodson's law, piecewise duration modulation make the changes in light and shadow parameters more in line with the psychological perception of color and duration, thereby improving the scientificity and effectiveness of emotion regulation. (3) By combining seat weight and emotional intensity priority arbitration strategy, the full healing effect of high priority seats can be guaranteed when the priority of multiple seats is significantly different, or visual interference can be reduced by saturation upper limit constraint and dynamic speed stagger when the priorities are similar, thus taking into account driving safety and overall healing benefits. Attached Figure Description

[0015] Figure 1 A flowchart illustrating the steps of a parallel emotion healing method based on emotion perception through multi-seat door projection. Figure 2 This is a schematic diagram of the projection module installation. Figure labeling: 1-Projection module, 2-Fabric area, 3-Handle block. Detailed Implementation

[0016] The following are specific embodiments of the present invention, which are described in conjunction with the accompanying drawings. However, the present invention is not limited to these embodiments.

[0017] In the current field of automotive cabin technology, solutions for environmental regulation based on occupant emotional states have seen some development. Existing technologies typically utilize in-vehicle cameras, microphones, or biosensors to collect occupants' facial expressions, voice, or physiological signals. These signals are then processed by emotion recognition models to output emotion categories or affective dimensions (such as valence and arousal), which in turn drive actuators such as ambient lighting, audio systems, air conditioning, or fragrances to produce corresponding sensory stimuli to improve or match the occupants' emotions. However, these solutions generally suffer from the following shortcomings: First, the adjustment methods are mostly limited to color gradation, brightness variation, or abstract lighting effects, lacking concrete dynamic lighting themes based on psychological and biological affinity design verification. Furthermore, the mapping from emotion to lighting parameters is mostly linear or follows fixed rules, failing to fully consider the non-linear characteristics of human visual perception and psychophysiological responses. Second, when faced with scenarios where multiple occupants coexist in the same cabin, existing technologies either employ a globally unified adjustment strategy or compromise the emotional needs of multiple occupants into a single instruction through weighted averaging or priority selection. This fails to provide truly personalized, non-interfering therapy for occupants in different seats simultaneously, and lacks a systematic solution to problems such as visual crosstalk and flickering when multiple lighting sources operate simultaneously. To address these technical issues, such as... Figure 1 As shown, this invention proposes a parallel emotion healing method based on multi-seat car door projection using emotion perception, comprising the following steps: S1. The emotional state of each seat occupant during the sliding window time is obtained through the cockpit perception system. The emotional state includes at least the average valence representing the degree of emotional pleasure and the average arousal representing the level of physiological activation. S2. Calculate the dynamic light and shadow parameters corresponding to each seat according to the emotional state of each seat occupant. The dynamic light and shadow parameters include at least the saturation based on the average valence and nonlinearly modulated according to Stevens' power law, the total adjustment time calculated piecewise according to Yerkes-Dodson's law based on the average arousal, and the dynamic speed corresponding to the emotional state. S3. Calculate the overall priority of each seat based on the safety priority of each seat and the emotional intensity obtained by weighting the normalized deviation calculated from the average valence and the average arousal. S4. Adjust the dynamic lighting parameters according to the ratio of the highest priority to the second highest priority in the overall priority. When the ratio is greater than or equal to the preset threshold, keep the seat corresponding to the highest priority running according to the original dynamic lighting parameters and control the other seats to execute the downgrade mode; otherwise, adjust the dynamic speed in the dynamic lighting parameters of each seat to be staggered and limit the saturation to within the preset upper limit. S5. Based on the adjusted dynamic light and shadow parameters, control the projection modules installed on the inside of each door to project dynamic light and shadow patterns onto the corresponding occupants.

[0018] The multi-seat door projection parallel emotion therapy method based on emotion perception provided by this invention can be implemented in various vehicle cabin configurations. Taking a five-seat passenger vehicle as an example, a cabin perception system is pre-installed inside the vehicle, mainly including a wide-angle cabin monitoring camera installed in the center of the ceiling or at the rearview mirror, with a field of view of no less than 120 degrees, capable of simultaneously covering the driver, front passenger, and the left and right rear seat areas. This camera has infrared supplementary lighting function to ensure clear acquisition of occupant facial images even at night or in low-light environments. Before the vehicle leaves the factory, engineers establish a three-dimensional spatial coordinate system for the cabin using calibration tools, with the vehicle center or steering wheel center as the origin, recording the exclusive spatial coordinate domain of each seat. This calibration data is stored in the non-volatile memory of the on-board controller for subsequent spatial weight filtering processing. Optionally, the system can also add a 60GHz millimeter-wave radar in the center of the vehicle ceiling for non-contact extraction of heart rate and respiratory rate to improve the reliability of arousal estimation.

[0019] During actual driving, the system acquires images from the camera at a fixed frequency and performs face detection and key point localization on each frame. Lightweight deep learning models such as MobileNet-SSD or RetinaFace can be used to output the bounding box of each face and the pixel coordinates of five key points: eyes, nose tip, and left and right corners of the mouth. Then, using the camera's intrinsic parameters (focal length, principal point coordinates, distortion coefficients) and a pre-calibrated 3D cockpit model, the pixel coordinates of each facial key point are converted into 3D spatial coordinates using a PnP algorithm or a direct regression method based on a deep neural network. To simplify computation, the center point of the face bounding box or the nose tip can be used as an approximate spatial location of the face.

[0020] After obtaining the 3D coordinates of each face, the system sequentially determines whether the coordinates fall within a pre-defined seat space. If the coordinates are within the calibrated space of a particular seat, the face is marked as an occupant of that seat, and the corresponding facial data weight is set to 100%, serving as valid data for subsequent emotion recognition. If the face coordinates do not fall within any seat calibrated space, such as in the aisle area between two seats, or clearly fall within the space of an adjacent seat (e.g., a face detected by the driver's side camera in the passenger seat), the weight of that face is directly set to zero, and the data is completely discarded, not participating in emotion calculation. For faces in the critical area of ​​the seat boundary, single-frame judgment may produce ambiguity. The system continuously tracks the movement trajectory of the face in several consecutive frames using a Kalman filter or a simple centroid tracking method. If a face briefly enters the critical area and then quickly leaves the seat space, it is marked as invalid data and the corresponding consecutive frames are discarded; if the face remains stably within the seat space, even if it is at the edge of the critical area, it is determined to be an occupant of that seat, and the weight is restored to 100%. Using the spatial weighted filtering algorithm described above, a single wide-angle camera can simultaneously separate the facial information of the driver, front passenger, and multiple rear passengers, avoiding the cost and system complexity of installing a separate camera for each seat.

[0021] After determining seat assignment, the system independently runs an emotion recognition process for each seat. Taking the driver's seat as an example, the facial image belonging to the driver is input into a multimodal emotion recognition model. This model is typically built on a deep neural network and can extract facial expression features from facial images, such as judging emotional tendencies through facial action unit analysis and head posture estimation. Simultaneously, the system can also utilize remote photoplethysmography (LPPG) technology to extract heart rate and heart rate variability indicators from facial videos, i.e., by capturing subtle color changes on the skin surface caused by the heart's pumping action, combined with signal processing algorithms to achieve non-contact heart rate estimation. When millimeter-wave radar is configured, the radar detects the micro-displacement of the occupant's chest cavity using frequency-modulated continuous wave technology, thereby extracting respiratory rate and heart rate. After fusing the above multi-source signals, the emotional state of each occupant is output. In this invention, the emotional state is represented using an affective dimension model, specifically including valence representing the degree of emotional pleasure and arousal representing the level of physiological activation. The valence range is typically set to [-1, 1], with negative values ​​indicating unpleasantness and positive values ​​indicating pleasure; the arousal range is set to [0, 1], with larger values ​​indicating higher levels of physiological activation. The system continuously acquires physiological signals and facial images of each seat using a fixed-length sliding window (e.g., a window length of 30 seconds, sliding once every 10 seconds), and outputs the average valence within the window. and average wakefulness This design ensures the stability of emotion recognition and avoids false triggers caused by single-frame noise or instantaneous facial expression fluctuations.

[0022] When the emotional state displayed in the sliding window for a particular seat meets a preset trigger condition, the system begins to adjust the lighting and shadows for that seat. In one embodiment of the invention, the trigger condition is set as follows: the emotional category is anger, sadness, or fatigue, and the average arousal level is... or average valence Or the emotion category is pleasant and the average arousal level is... or average valence For the three negative emotions of anger, sadness, and fatigue, the regulation goal is to guide the emotion towards a calmer state. It should be noted that the calmer state is defined in this invention as the average valence. And average wakefulness The benchmark for calmness is set at 0.5 instead of the neutral point of 0, based on the following considerations: A valence of 0 corresponds to a neutral state in the emotional dimension model, often manifesting as boredom or emotional poverty, which is not an ideal therapeutic endpoint; while true calmness and tranquility is a low-arousal state with positive valence, where the individual can subjectively experience significant relaxation and satisfaction. For pleasurable emotions, the regulatory goal is to maintain stable or moderately decrease arousal levels while preserving positive valence, preventing emotional fatigue or distraction resulting from over-excitement.

[0023] Once regulation is triggered, the system immediately adjusts the average valence output by the sliding window. and average wakefulness The saturation and total duration of this adjustment are calculated in one step. The calculation of saturation strictly follows Stevens' power law. Let the baseline valence be... Set to 0.5, baseline saturation. Take 30%, maximum saturation Take 100%. Valence deviation (i.e., when in a negative emotional state) Maximum valence deviation The value is 1.5 (because the minimum valence is -1, and the difference between it and 0.5 is 1.5), so the saturation... It can be determined by the following formula: , in, is the power exponent of Stevens' power law. When At that time, regardless of the specific value of the valence, the saturation is fixed at the baseline saturation. The physical significance of this nonlinear mapping lies in simulating the diminishing marginal returns of human perception of color saturation: in low-saturation regions, the human eye is more sensitive to color changes, and appropriately increasing saturation can significantly alter visual perception; while in high-saturation regions, the sensitivity of the human eye decreases, and the perceptual gain from further increasing saturation gradually diminishes. Using a power function allows saturation to increase rapidly at first and then slowly as valence deviation increases, which is more in line with psychophysical principles, thereby improving the accuracy of emotion regulation and user experience.

[0024] The calculation of the total adjustment time is based on the Yerkes-Dodson law, using a piecewise power function model, with the minimum adjustment time set as... The adjustment time corresponding to moderate arousal is The maximum adjustment time is The optimal arousal level is .

[0025] If the detected average wakefulness Then the low wake-up area formula is used: , in For an adjustable low wake-up exponent, like Then the high wake-up area formula is used: , in It is an adjustable power exponent for the high wake-up zone.

[0026] The rationale behind this segmented design is that when occupants are in a low-arousal state (such as fatigue or drowsiness), moderate light and shadow stimulation can quickly increase arousal, thus the adjustment time decreases as the initial arousal increases. However, when occupants are in a high-arousal state (such as anger or fear), physiological and psychological calming requires a longer time, and the required time increases exponentially with increasing arousal. This is achieved by introducing… Power functions greater than 1 can more realistically reflect the non-linear delay characteristics of emotional calming in a high-arousal state. Brightness values ​​are pre-calibrated to fixed values ​​by emotion category, with different percentages assigned to different emotion categories. These fixed-matching brightness values ​​can be proportionally increased when ambient light is strong, or appropriately decreased when interior reflections are significant, but they do not change continuously with valence or arousal level.

[0027] After calculating the independent dynamic lighting parameters for each seat, the system also needs to address resource conflicts in multi-occupant parallel scenarios. Due to limited cabin space, even with spatial isolation design for the projection modules corresponding to each seat, visual interference may still occur when multiple dynamic lighting effects run simultaneously. Therefore, this invention introduces a dual-mode conflict resolution strategy based on comprehensive priority. The system first calculates the comprehensive priority of each seat. The comprehensive priority consists of a product of two parts: the seat's preset safety priority weight and the emotional intensity. Seat safety priority weight... This reflects the paramount importance of driving safety, exemplarily setting the driver's seat weight to 3, the front passenger seat weight to 2, and the rear seats weight to 1. Emotional intensity is obtained by a weighted combination of normalized valence deviation and arousal, where the normalized valence deviation of the i-th seat... It reflects the degree to which the occupants' emotions deviate from a calm baseline, among which Let be the average valence of the i-th seat. Then, combine the valence deviation with the average arousal of the corresponding seat. Assign weight coefficients respectively and ,and The emotional intensity of that seat is... The overall priority of the i-th seat. The formula takes into account both the safety importance of the seat and the urgency of the passenger's current emotional state.

[0028] After obtaining the overall priority of each seat, the system identifies the highest priority. and second highest priority And compare the ratio of the two. If (in If the threshold is set to a preset value, it is determined to be mode A (single high priority); otherwise, it is determined to be mode B (similar priority).

[0029] In Mode A, the system determines that a particular seat has a significantly higher emotional need than other seats and should be prioritized for its therapeutic effect. In practice, this high-priority seat operates exactly as calculated in step S2, with the original saturation, brightness, and duration, without any reduction. All other seats enter a downgrade mode: their saturation is reduced to 50% of the current valence-corresponding saturation, but not below the baseline saturation (30%); simultaneously, their brightness is reduced to 50% of the preset fixed brightness value corresponding to the seat's emotional category. This preserves the basic therapeutic function of other seats while significantly reducing their visual interference with the high-priority seat occupant's field of vision.

[0030] In Mode B, the emotional needs of each seat are not significantly different, and it is neither possible nor appropriate to simply downgrade one side. In this case, the system adopts an alternative strategy: imposing interference suppression constraints on all seats involved in the conflict. First, the saturation of all seats is limited to a preset upper limit, for example, 70%. If the originally calculated saturation of a seat is already below this upper limit, it remains unchanged; if it is above the upper limit, it is forcibly reduced to 70%.

[0031] Secondly, the system detects the dynamic cycle of any two seats (i.e., the repetition cycle of the main action in each light and shadow theme, such as the interval of water ripple diffusion). If the ratio of the dynamic cycles of the two seats falls within a preset range, such as 0.8 to 1.25, it indicates that the movement speeds of the two lights and shadows are too close, easily creating a sense of synchronization or overlapping flickering. In this case, the system multiplies the cycle of one seat by a factor, such as multiplying by 1.3 or dividing by 1.3, so that the ratio of the two cycles falls outside the aforementioned range. The adjustment range usually does not exceed 30% of the original cycle to ensure that the adjusted cycle still falls within the effective speed range of the emotional theme and does not disrupt the original healing rhythm. Through the above-mentioned saturation upper limit constraint and dynamic speed staggering, the system reduces visual conflict when multiple lights and shadows are in parallel without significantly sacrificing individual healing effects.

[0032] After determining the final dynamic lighting parameters (saturation, brightness, duration, speed) for each seat and whether to perform downgrading or speed adjustment, the system begins to control the projection modules installed on the inside of each door. For example... Figure 2As shown, the projection module 1 for each door is hidden below the front end of the door armrest (i.e., the bottom of the handle block 3), completely invisible from the outside. The light outlet of the projection module 1 faces upwards at an angle, projecting light and shadow onto the fabric area 2 in the upper middle part of the door panel (i.e., the dynamic light and shadow pattern projection area). This fabric area 2 is made of a material with diffuse reflection properties, such as light gray recycled polyester fiber or fabric-textured leather, which can evenly scatter the obliquely incident projection light towards the occupants, avoiding glare. The projection module 1 contains an RGB three-color LED light source, a focusing and imaging lens group, and a pattern generating element. The pattern generating element can be a film or a microlens array, pre-made with natural biomimetic patterns such as water ripples, starlight, and dawn light. To produce dynamic effects, the central controller sends an LED lighting sequence according to a preset timing sequence, such as water ripples lighting up in concentric circles from the center outwards, or starlight flowing along a wave-shaped trajectory. The driving chip works in conjunction with the static pattern element to produce a smooth animation effect. All projection modules 1 are connected to the cockpit main control via LIN or CAN bus, receiving control commands for theme switching, saturation, brightness, speed, etc. Furthermore, to further suppress light crosstalk between multiple door projections, each projection module 1's light outlet is equipped with a light-shielding grille, confining the main energy of the projected beam to the projection area of ​​its own door trim panel, preventing direct illumination of the center console, opposite door panels, or other seating areas. Simultaneously, the door panel fabric surface is coated with a directional diffuse reflection coating, causing reflected light to primarily point towards the side view of the corresponding occupant, significantly attenuating the intensity of scattered light towards the center of the cabin. Through the combination of these optical hardware measures and software conflict resolution strategies, each occupant receives personalized, uninterrupted therapeutic lighting at the same time.

[0033] After each projection adjustment, the system also includes a parameter adaptive adjustment mechanism based on effect feedback. Specifically, after the adjustment is complete, the system restores the default ambient lighting and re-collects the average valence and average arousal of each seat within a sliding window of the same length (e.g., 30 seconds), denoted as follows: and According to the pre-adjustment , With adjustment The system calculates a continuous reward value to evaluate the effectiveness of this adjustment.

[0034] For negative emotions such as anger, sadness, and fatigue, define the valence increase. Affecting wakefulness Let the maximum effective improvement be defined. If the value is 0.5, then the normalized valence improvement is... Normalized arousal level improved Reward Value ,in This is the adjustable weighting coefficient for negative emotions. If the adjustment meets the success criteria (i.e....) and If so, an additional adjustable bonus value will be added. However, the maximum reward value is 1.

[0035] Here, the goal of modulating negative emotions using arousal reduction is to guide passengers from a state of high arousal to a calm, low arousal zone. Therefore, the system should clearly reward a decrease in wakefulness and punish an increase in wakefulness. A positive value indicates that the arousal level has been successfully reduced (good effect). A negative value indicates that arousal level actually increases (poor performance). Therefore, in the reward function, through... Only positive values ​​are retained, so that the reward value is only awarded for effective reduction results, and not for increase results.

[0036] For pleasurable emotions, the valence increase was also defined. The difference is that here, the absolute value of the change in arousal level is defined separately. Reward value ,in , , and These are adjustable weighting coefficients for valence and arousal stability, respectively. Through the aforementioned continuous reward function, the system can more precisely distinguish the effects of different levels of improvement, rather than simply relying on a binary success / failure judgment.

[0037] The goal of regulating pleasurable emotions using the absolute value of arousal changes is to maintain the current positive state and avoid over-excitement or a sudden drop in mood. The system does not expect arousal to unidirectionally increase or decrease, but rather to remain stable. Therefore, the reward function uses... By retaining stable terms, the reward value is increased to reward stable results and decreased to reward unstable results.

[0038] Based on this, the system employs an ε-greedy strategy and Q-learning to perform personalized optimization of the adjustable parameters. The adjustable parameters include the minimum adjustment time. Medium adjustment duration Maximum adjustment duration Low wake-up zone power index High Arousal Zone Power Index Optimal wakefulness The system maintains an independent parameter-value table (Q-table) for each user, recording the cumulative reward expected for performing a parameter adjustment action (such as increasing, decreasing, or keeping it unchanged) under different states. In each adjustment interaction, the system selects the action with the highest current Q-value with a high probability, and randomly selects an action with a lower probability. The latter aims to prevent the system from getting stuck in local optima, while the former ensures that existing experience is used in most cases. After adjustment, the Q-value is updated based on the calculated reward value R. To avoid frequent parameter changes due to single random fluctuations, the system also introduces a conservative switching mechanism: the parameter adjustment corresponding to an action is only applied to the system when the Q-value of a certain action exceeds a preset threshold of the Q-value of the current optimal action in multiple consecutive interactions (e.g., 5 consecutive times). If this condition is not met, even if the Q-value increases in a single instance, the parameter is not switched immediately. New users' initial parameters use the default values ​​calibrated during the pre-research phase. With user consent and anonymization, the system can upload the converged parameter vector after each successful interaction to the cloud. The cloud server aggregates parameters from a large number of users, periodically calculates the weighted average of each parameter, and uses this to update the default parameter library and distribute it to vehicles via OTA. All users' personal physiological data, adjustment history, and personalized parameters are stored only in the local vehicle controller. Users can clear historical data or restore factory settings at any time.

[0039] In summary, this invention, by equipping each seat with an independent door projection module and combining it with a spatial weighted filtering algorithm, achieves crosstalk-free parallel acquisition and independent healing of the emotional states of multiple occupants under a single camera. This allows occupants in different seats to receive personalized light and shadow interventions without interference at the same time. Furthermore, based on Stevens' power law-based nonlinear saturation modulation and Yerkes-Dodson's law-based segmented duration modulation, the changes in light and shadow parameters better align with human psychological perception of color and duration, thereby enhancing the scientific rigor and effectiveness of emotion regulation. Simultaneously, by employing a priority arbitration strategy that integrates seat weight and emotional intensity, the system can ensure the complete healing effect of high-priority seats when there are significant differences in priority among multiple seats, or reduce visual interference when priorities are similar by constraining saturation limits and dynamically staggering speeds, balancing driving safety and overall healing benefits. Moreover, a reinforcement learning closed-loop feedback mechanism built upon changes in valence and arousal before and after adjustment enables the system to continuously optimize modulation parameters for the personalized responses of different users, gradually approaching a better healing effect over long-term use.

[0040] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture. If the specific posture changes, the directional indication will also change accordingly.

[0041] Furthermore, in this invention, descriptions involving terms such as "first," "second," and "a" are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0042] In this invention, unless otherwise explicitly specified and limited, the terms "connection," "fixed," etc., should be interpreted broadly. For example, "fixed" can mean a fixed connection, a detachable connection, or an integral part; it can mean a mechanical connection or an electrical connection; it can mean a direct connection or an indirect connection through an intermediate medium; it can mean the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0043] Furthermore, the technical solutions of the various embodiments of the present invention can be combined with each other, but only if they are feasible for those skilled in the art. If the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

Claims

1. A parallel emotional therapy method based on multi-seat car door projection using emotion perception, characterized in that, Including the following steps: S1. The emotional state of each seat occupant during the sliding window time is obtained through the cockpit perception system. The emotional state includes at least the average valence representing the degree of emotional pleasure and the average arousal representing the level of physiological activation. S2. Calculate the dynamic light and shadow parameters corresponding to each seat according to the emotional state of each passenger. The dynamic light and shadow parameters include at least the saturation based on the average valence and nonlinearly modulated according to Stevens' power law, the total accommodation time calculated piecewise according to Yerkes-Dodson's law based on the average arousal, and the dynamic speed corresponding to the emotional state. S3. Calculate the overall priority of each seat based on the safety priority of each seat and the emotional intensity obtained by weighting the normalized deviation calculated from the average valence and the average arousal. S4. Adjust the dynamic lighting parameters according to the ratio of the highest priority to the second highest priority in the overall priority. When the ratio is greater than or equal to the preset threshold, keep the seat corresponding to the highest priority running according to the original dynamic lighting parameters and control the other seats to execute the downgrade mode; otherwise, adjust the dynamic speed in the dynamic lighting parameters of each seat to be staggered and limit the saturation to within the preset upper limit. S5. Based on the adjusted dynamic light and shadow parameters, control the projection modules installed on the inside of each door to project dynamic light and shadow patterns onto the corresponding occupants.

2. The method for parallel emotional healing based on multi-seat car door projection using emotion perception as described in claim 1, characterized in that, In step S1, the cockpit perception system includes a wide-angle cockpit monitoring camera. Through spatial three-dimensional calibration and spatial weight filtering algorithm, the detected three-dimensional coordinates of the face are compared with the preset seat space domain. If the three-dimensional coordinates of the face fall into a certain seat space domain, the facial data of the face is marked as a valid input of the seat and used for emotion recognition of the seat. If the three-dimensional coordinates of a face do not fall within any seat space domain or fall within an adjacent seat space domain, the facial data of that face is marked as invalid and discarded.

3. The method for parallel emotional healing based on multi-seat car door projection using emotion perception as described in claim 1, characterized in that, In step S2, the saturation based on the average valence nonlinearly modulated according to Stevens' power law is expressed by the following formula: In the formula, The average valence detected within the sliding window. The saturation level corresponding to the average valence. As the baseline saturation, For maximum saturation, Valence deviation, For calm baseline valence, For the maximum valence deviation, Let be the power exponent of Stevens' power law, and when The saturation is fixed at time. .

4. The method for parallel emotional healing based on multi-seat car door projection using emotion perception as described in claim 3, characterized in that, In step S2, the total adjustment time, calculated piecewise according to the Yerkes-Dodson Law based on the average arousal level, is expressed by the following formula: In the formula, The average wakefulness detected within the sliding window. The total settling time corresponding to the average arousal level. For optimal wakefulness, These represent the minimum, medium, and maximum adjustment durations, respectively. The power exponent for low wake-up region. It is a power exponent of the high wake-up region, and .

5. The method for parallel emotion therapy based on multi-seat car door projection using emotion perception as described in claim 4, characterized in that, In step S3, the overall priority of each seat is calculated, expressed by the formula: In the formula, The overall priority of the i-th seat. The preset safety priority weight for the i-th seat Let i be the normalized valence deviation of the i-th seat. Let i be the average valence of the i-th seat. Let i be the average arousal level of the i-th seat. These are the weighting coefficients, and .

6. The method for parallel emotion therapy based on multi-seat car door projection using emotion perception as described in claim 5, characterized in that, Following step S5, the following step is also included: The average valence and average arousal of the crew were re-collected, and the continuous reward value was calculated based on the changes before and after adjustment. The minimum adjustment duration was updated using an ε-greedy strategy and Q-learning. Medium adjustment duration Maximum adjustment duration Low wake-up zone power index High Arousal Zone Power Index Optimal wakefulness One or more of them.

7. The method for parallel emotional healing based on multi-seat car door projection using emotion perception as described in claim 1, characterized in that, In step S4, the downgrade mode is specifically as follows: Using the baseline saturation as the minimum adjustment limit for saturation, the saturation of the remaining seats is adjusted to 50% of the saturation corresponding to the average valence, and the brightness of the remaining seats is reduced to 50% of the preset fixed brightness value corresponding to the mood category of that seat.

8. The method for parallel emotional healing based on multi-seat car door projection using emotion perception as described in claim 1, characterized in that, In step S4, the dynamic speeds of each seat are adjusted to be staggered, specifically as follows: Detect the ratio of the dynamic cycles of any two seats. If the ratio falls within a preset range, multiply the dynamic cycle of one of the seats by a preset multiplier factor.

9. The method for parallel emotional healing based on multi-seat car door projection using emotion perception as described in claim 1, characterized in that, In step S5, a light-shielding grille is installed at the light outlet of the projection module. The light-shielding grille is used to limit the horizontal divergence angle of the dynamic light and shadow pattern within a preset angle range.

10. The method for parallel emotion therapy based on multi-seat car door projection using emotion perception as described in claim 1, characterized in that, The vehicle door is equipped with a dynamic light and shadow pattern projection area and coated with a directional diffuse reflection layer, which reflects the dynamic light and shadow pattern to the lateral field of vision of the corresponding occupant through directional diffuse reflection.

Images