Driver spinal health monitoring method, system and automotive intelligent cockpit

Abstract

This invention provides a method, system, and intelligent cockpit for monitoring driver spinal health, comprising: collecting physiological data of the driver through onboard multi-source sensors; preprocessing and extracting features from the physiological data to obtain the driver's characteristic parameters; and retrieving the driver's spinal health indicators based on the characteristic parameters and a pre-built driver spinal health database. This invention can provide drivers with personalized quantitative assessments of spinal health, improving the accuracy of driver spinal health monitoring.

Description

Technical Field

[0001] This invention relates to the field of automotive intelligent cockpit technology, and in particular to a method, system and intelligent cockpit for monitoring driver spinal health. Background Technology

[0002] Most current health chairs or monitoring solutions on the market are based on a single type of sensor (such as pressure distribution pads or heart rate belts) or simple rules (such as sedentary timing). These technologies can only provide basic posture abnormality warnings or fatigue alerts, lacking the ability to deeply model the biomechanical load on the spine and thus unable to quantitatively assess and predict spinal health risks. Summary of the Invention

[0003] In view of this, the purpose of the present invention is to provide a method, system and intelligent cockpit for monitoring the spinal health of drivers, which can provide drivers with personalized quantitative assessment of spinal health and improve the accuracy of monitoring the spinal health of drivers.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows: In a first aspect, the present invention provides a method for monitoring the spinal health of a driver, comprising: collecting physiological data of the driver through an on-board multi-source sensor; preprocessing and extracting features from the physiological data to obtain the driver's feature parameters; and retrieving the driver's spinal health indicators based on the feature parameters and a pre-constructed driver spinal health database.

[0005] Optionally, the feature parameters include at least: height, weight, trunk lean angle, center of gravity, and steering wheel grip strength; based on the feature parameters and a pre-built driver spinal health database, the driver's spinal health indicators are retrieved, including: based on the trunk lean angle, a first-level search is performed in the pre-built driver spinal health database to obtain a first candidate sample; a weighted feature vector of the feature parameters is calculated based on pre-set weight coefficients, and based on the weighted feature vector, a second-level search is performed in the first candidate sample to obtain a preset number of second candidate samples; the driver's spinal health indicators are calculated based on the spinal health indicators of the second candidate samples.

[0006] Optionally, based on the trunk forward tilt angle, a first-level search is performed in a pre-built driver spinal health database to obtain the first candidate sample, including: calculating the difference between the trunk forward tilt angle and the trunk forward tilt angle of each sample in the driver spinal health database; determining samples whose absolute value of the difference is less than or equal to a preset value as candidate samples; and filtering the candidate samples based on height and weight to obtain the first candidate sample.

[0007] Optionally, based on the weighted feature vector, a secondary search is performed in the first candidate sample to obtain a preset number of second candidate samples, including: calculating the similarity between the weighted feature vector and the weighted feature vector of each sample in the first candidate sample; and selecting a preset number of first candidate samples as second candidate samples based on the similarity.

[0008] Optionally, the spinal health index of the driver can be calculated based on the spinal health index of the second candidate sample, including: weighting the spinal health index of the second candidate sample based on similarity to obtain the spinal health index of the driver.

[0009] Optionally, after retrieving the driver's spinal health index based on feature parameters and a pre-built driver spinal health database, the method further includes: determining the driver's spinal health score based on the driver's spinal health index and preset scoring rules; generating the driver's core muscle load status and prompt information based on the muscle activation of the second candidate sample; and displaying the driver's spinal health score, core muscle load status, and prompt information on the vehicle's infotainment screen.

[0010] Optionally, before collecting the driver's physiological data through onboard multi-source sensors, the process includes: collecting the physiological data of the subjects and extracting key feature parameters from the physiological data; wherein the key feature parameters include at least: height, weight, joint length, trunk lean angle, center of gravity, steering wheel grip strength, and spinal health status; performing simulations based on the key feature parameters of the subjects to obtain the subjects' spinal health indicators and muscle activation; constructing a driver spinal health database based on the subjects' key feature parameters, spinal health indicators, and muscle activation, and determining the weight coefficients of the key feature parameters.

[0011] Secondly, the present invention provides a driver spinal health monitoring system, comprising: a data acquisition module for acquiring physiological data of the driver through an on-board multi-source sensor; a feature extraction module for preprocessing and extracting features from the physiological data to obtain the driver's feature parameters; and a retrieval module for retrieving the driver's spinal health indicators based on the feature parameters and a pre-built driver spinal health database.

[0012] Thirdly, the present invention provides an intelligent cockpit for automobiles, including a processor and a memory, the memory storing computer-executable instructions that can be executed by the processor, the processor executing the computer-executable instructions to implement the steps of the method provided in any of the first aspects above.

[0013] Fourthly, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, performs the steps of the method provided in any of the first aspects above.

[0014] This invention brings the following beneficial effects: The driver spinal health monitoring method, system, and intelligent cockpit provided by this invention first collect physiological data of the driver through in-vehicle multi-source sensors; then, the physiological data is preprocessed and features are extracted to obtain the driver's feature parameters; finally, the driver's spinal health indicators are obtained by searching based on the feature parameters and a pre-built driver spinal health database. In this method, by fusing in-vehicle multi-source sensors, it can comprehensively and synchronously capture multi-dimensional key physiological data that determines spinal stress, such as body posture, pressure distribution, and upper limb load, thus improving the problem of single data dimensions in existing technologies. By searching through the pre-built driver spinal health database and the extracted multi-source feature parameters to obtain the driver's spinal health indicators, it can provide drivers with personalized quantitative assessments of spinal health, thereby improving the accuracy of driver spinal health monitoring.

[0015] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention are realized and obtained in accordance with the structures particularly pointed out in the description, claims and drawings.

[0016] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0017] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 A flowchart of a method for monitoring the spinal health of a driver provided in an embodiment of the present invention; Figure 2 A flowchart of experimental data acquisition is provided for an embodiment of the present invention; Figure 3 A flowchart for spinal health assessment provided in this embodiment of the invention; Figure 4 A schematic diagram of an HMI interface provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of a method for monitoring the spinal health of a driver provided in an embodiment of the present invention; Figure 6 This is a structural schematic diagram of an intelligent cockpit for automobiles provided in an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Currently, most common health chairs or monitoring solutions are based on a single type of sensor (such as pressure distribution pads or heart rate belts) or simple rules (such as sedentary timing), which have problems such as limited data dimensions, lack of personalized monitoring results, low accuracy, and inability to quantitatively assess and predict spinal health risks.

[0021] Based on this, the driver spinal health monitoring method, system and intelligent cockpit provided by the embodiments of the present invention can provide drivers with personalized quantitative assessment of spinal health, thereby improving the accuracy of driver spinal health monitoring.

[0022] To facilitate understanding of this embodiment, a method for monitoring driver spinal health disclosed in this invention will first be described in detail. This method can be executed by the processor of a car's intelligent cockpit. See also... Figure 1 The flowchart shown illustrates a method for monitoring the spinal health of a driver, indicating that the method mainly includes the following steps S101 to S103: Step S101: Collect the driver's physiological data through on-board multi-source sensors.

[0023] In one implementation, the driver's physiological data includes, but is not limited to: human driving posture, gunshot height, weight, real-time body pressure distribution, and steering wheel grip strength. Among them, (1) the driver's image is collected using a camera or millimeter-wave radar installed in the vehicle (such as the A-pillar or rearview mirror), and the human driving posture is identified based on the collected image, and the torso forward tilt angle and pelvic tilt angle are estimated in real time (defined in the same way as Xsens). The specific operation is as follows: 1) the driver's body contour is extracted by image segmentation technology; 2) key point detection algorithm is used to identify key points such as the top of the head, neck, shoulders, and waist, and the spinal curve is formed by connecting the key points; 3) the bending angle of the spinal curve in different segments is calculated, including parameters such as torso forward tilt angle and pelvic tilt angle.

[0024] (2) The driver's sitting height is inferred from the stroke encoder of the seat lift motor, and the driver's height is estimated by combining the human body proportion model. The specific operation is as follows: 1) Read the pulse count or absolute position value of the seat lift motor encoder. P current And to obtain the mechanical parameters of the seat (encoder values ​​of the seat's mechanical zero point (usually the lowest point or design reference point)). P zero And transmission ratio and lead coefficient of the lead screw K gear ), and calculate the vertical lift Δ of the seat relative to the zero position. H motor =( P current P zero )× K gear 2) The vertical distance from point H (hip joint center) to the horizontal plane where the accelerator pedal heel is located. H base The vertical lift Δ of the seat relative to the zero position H motor Make adjustments to obtain an effective seating height; for simple vertically adjustable seats: H 30 measured ≈ H base +Δ H motor ×cos( θ ),in, θ For seat tilt angle; for multi-link mechanisms: using a preset lookup table or polynomial fitting function, map the motor travel to the actual Z-axis coordinate of point H. Z H point_reference Thus, an effective sitting height is achieved. H 30 measured = Z H point_reference + f (Δ H motor 3) After obtaining the accurate H30, it is necessary to establish the relationship between sitting height and body height using anthropometry data. A linear regression model can be used: Lstature = α × H 30 measured + β + ;in, α The value ranges from 1.6 to 1.8. βUsed to compensate for differences in trunk length and pelvic thickness. The residuals represent individual differences; 4) The calculated H 30 measured Substitute the cumulative distribution function of the human body size database to determine the percentile of the population corresponding to the H30 value, and take the average height corresponding to the percentile as the height estimate.

[0025] (3) The driver’s weight is collected by a weighing sensor (±0.5kg) integrated under the seat.

[0026] (4) Pressure magnitude and distribution data are collected in real time through a high-resolution thin-film pressure distribution sensor array embedded in the headrest, seat cushion and backrest.

[0027] (5) The driver’s grip strength and distribution are monitored by a miniature pressure or strain sensor integrated in the steering wheel rim.

[0028] Step S102: Preprocess and extract features from the physiological data to obtain the driver's feature parameters.

[0029] In one implementation, a Kalman filter (the state vector includes posture angle, center of gravity, and grip strength) is used to filter noise from the physiological data. The parameters of the Kalman filter can be fixed according to the sensor characteristics. Based on the filtered data, feature parameters such as height, weight, torso lean angle, center of gravity, and steering wheel grip strength are extracted, and these feature parameters are aligned to obtain a feature vector: [ H , W , A ,( x c , y c ), F ];in, H For height, W For weight, A To reduce the forward tilt, ( x c , y c () represents the coordinates of the center of gravity of the pressure. F Steering wheel grip strength.

[0030] In this embodiment of the invention, feature alignment ensures that the definitions, units, and normalization methods of the torso lean angle, center of gravity, and steering wheel grip force extracted from the vehicle are completely consistent with the data in the pre-built driver spinal health database, thereby ensuring the accuracy of subsequent database queries.

[0031] Step S103: Based on the feature parameters and the pre-built driver spinal health database, retrieve the driver's spinal health indicators.

[0032] In one implementation, based on feature parameters and a pre-built driver spinal health database, a two-level matching algorithm combining fast coarse screening based on B+ tree index and precise retrieval based on weighted Mahalanobis distance is used to screen multiple candidate samples, and the driver's spinal health index is determined based on the spinal health index of the candidate samples; wherein, the spinal health index can be L4 / L5 intervertebral disc pressure and muscle activation.

[0033] The driver spinal health monitoring method provided in this invention integrates multiple in-vehicle sensors to comprehensively and synchronously capture multi-dimensional key physiological data such as body posture, pressure distribution, and upper limb load that determine spinal force, thus improving the problem of single data dimension in the prior art. By retrieving data from a pre-built driver spinal health database, the driver's spinal health indicators can be obtained, providing the driver with a personalized quantitative assessment of spinal health, thereby improving the accuracy of driver spinal health monitoring.

[0034] In one implementation, the characteristic parameters include at least: height, weight, torso lean angle, center of gravity, and steering wheel grip strength. For the aforementioned step S103, i.e., retrieving the driver's spinal health indicators based on the characteristic parameters and a pre-built driver spinal health database, the following methods, including but not limited to, can be used, specifically including the following processes: (1) Based on the forward tilt angle of the trunk, a first-level search is performed in the pre-constructed driver spinal health database to obtain the first candidate sample.

[0035] In one implementation, firstly, the difference between the torso lean angle and the torso lean angle of each sample in the driver's spinal health database is calculated; then, samples whose absolute value of the difference is less than or equal to a preset value are identified as candidate samples; finally, the candidate samples are screened based on height and weight to obtain the first candidate sample.

[0036] In practice, the forward tilt angle of the torso is used as the basis for this. A Using the primary key, a B+ tree index is created for the database. The internal nodes (non-leaf nodes) of this index only store the index key value (i.e., a specific numerical value for the torso lean angle, such as 5°, 10°, or 15°) and pointers to child nodes; they do not store specific driver record details. Leaf nodes store all index key values ​​(…). A (and the corresponding data row pointers.) All records are strictly based on the torso tilt angle. A Arrange in ascending order. If there are multiple records... A If values ​​are the same (e.g., multiple times are both 15°), they are sorted sequentially within the leaf nodes according to a secondary sort key (e.g., timestamp). When a node is full, a split operation is triggered to maintain the balance of the tree and ensure that the lookup path length for any data is consistent.

[0037] Specifically, filter to meet | A current A db Samples with a | ≤ 5° were identified as candidate samples; among them, A current The current torso lean angle of the driver. A db The trunk forward tilt angle for each sample in the driver spinal health database is preset to 5°.

[0038] To further improve matching accuracy, this embodiment of the invention can also use height and weight for filtering, controlling the size of the first candidate sample to ≤500 entries and ensuring matching time <100ms. Specifically, height ±5cm and weight ±5kg can be added to the candidate samples for filtering, that is, selecting candidate samples whose height meets the driver's height ±5cm and driver's weight ±5kg. For example, if the driver's height is 175cm and weight is 70kg, then the first candidate sample is selected from the candidate samples with height in the range [170cm, 180cm] and weight in the range [65kg, 75kg].

[0039] (2) Calculate the weighted feature vector of the feature parameters based on the pre-set weight coefficients, and perform secondary retrieval in the first candidate sample based on the weighted feature vector to obtain a preset number of second candidate samples.

[0040] In one implementation, firstly, the similarity between the weighted feature vector and the weighted feature vector of each sample in the first candidate sample is calculated; then, based on the similarity, a preset number of first candidate samples are selected as second candidate samples.

[0041] In practical implementation, for the first candidate sample obtained from the aforementioned steps, firstly, a weighted feature vector of the feature parameters is calculated based on pre-set weight coefficients, i.e. ,in, , , , , These are the weight coefficients for the corresponding feature parameters; then, the similarity between the weighted feature vector and the weighted feature vectors of each sample in the first candidate sample is calculated using Mahalanobis distance, as shown in the following formula:

[0042] in, This is the driver's weighted feature vector. Σ is the weighted feature vector of each sample in the first candidate sample, and Σ is the feature covariance matrix of the training set, which is calculated offline by the database and fixed to the vehicle system.

[0043] Then, the samples in the first candidate sample are sorted in descending order of similarity, and the first candidate sample of the first preset number (e.g., K=5) is selected as the second candidate sample.

[0044] (3) The spinal health index of the driver is calculated based on the spinal health index of the second candidate sample.

[0045] In one implementation, the spinal health indicators of the second candidate samples are weighted and averaged based on similarity to obtain the driver's spinal health indicator. Specifically, the driver's spinal health indicator is obtained by weighting and averaging the spinal health indicators (such as L4 / L5 intervertebral disc pressure) corresponding to the K most similar second candidate samples. The formula is as follows:

[0046] In one implementation, after retrieving the driver's spinal health indicators based on feature parameters and a pre-built driver spinal health database, the method further includes: determining the driver's spinal health score based on the driver's spinal health indicators and preset scoring rules; generating the driver's core muscle load status and prompt information based on the muscle activation of a second candidate sample; and displaying the driver's spinal health score, core muscle load status, and prompt information on the vehicle's infotainment screen.

[0047] In practice, spinal health indicators can be mapped to a 0-100 score according to a preset scoring rule. The preset scoring rule could be: when the spinal health indicator is less than the first threshold, the corresponding score is 90-100 (Excellent); when the spinal health indicator is greater than or equal to the first threshold and less than the second threshold, the corresponding score is 70-89 (Good); when the spinal health indicator is greater than or equal to the second threshold and less than the third threshold, the corresponding score is 50-69 (Moderate); and when the spinal health indicator is greater than the third threshold, the corresponding score is less than 50 (Concern). For example: <1000N scores 90-100 (Excellent). The value is 1000-1500N, and the rating is 70-89 (good). It has a voltage rating of 1500-2000N and a score of 50-69 (medium). >2000N, rating <50 (Follow).

[0048] Based on this, in this embodiment of the invention, after obtaining the driver's spinal health indicators, a spinal health score can be obtained according to the aforementioned preset scoring rules. Simultaneously, based on the muscle activation level of the second candidate sample, the driver's core muscle load status and prompt information are generated, and the driver's spinal health score, core muscle load status, and prompt information are displayed on the vehicle's infotainment screen. For example, based on the spinal health indicators, the following information is generated: 1) Real-time score (e.g., "78 points, moderate load"); 2) Muscle tension prompt (e.g., "High erector spinae activation"); 3) Posture suggestions (e.g., "Suggest increasing backrest tilt angle by 3°"). The above scores, core muscle load status, and text prompts are displayed on the vehicle's infotainment screen through the HMI interface for the driver's reference.

[0049] This invention also provides a method for constructing a driver spinal health database, which can be implemented in ways including but not limited to the following: (1) Collect physiological data of the subjects and extract key feature parameters from the physiological data; among them, the key feature parameters include at least: height, weight, joint length, trunk forward tilt angle, center of gravity of pressure, steering wheel grip strength, and spinal health status.

[0050] In practice, the following physiological data were collected from the subjects in a laboratory environment: Height, weight, and joint length: measured directly using an electronic scale (±0.1kg) and a steel tape measure (±1mm); Body pressure distribution: collected by a high-resolution thin-film pressure sensor array (such as Tekscan I-Scan, 16×16 resolution) embedded in the headrest, seat cushion and backrest, with a sampling rate of 100Hz; Human driving posture: The human driving posture was obtained by using the Xsens MVN Awinda inertial motion capture system (17 sensor nodes, sampling rate of 120Hz, angle accuracy of ±0.5°) through a three-step process of T-pose static calibration, dynamic walking calibration, and natural and comfortable sitting posture dynamic calibration in the driving simulator. Steering wheel grip force: collected by micro strain gauge sensors (4 points / hand) integrated inside the steering wheel rim, with a sampling rate of 100Hz; Spinal health status (spine health and non-health): For non-healthy status, participants must provide an MRI / CT diagnostic report issued by a tertiary hospital, clearly identifying the affected segment (such as L4 / L5 disc herniation, degeneration grade, etc.), and the data must be anonymized.

[0051] Furthermore, key feature parameters are extracted, including: Height H (cm), weight W (kg), joint length: direct measurements; Trunk anteversion angle A (°): The angle between the T8 thoracic vertebra and the sagittal plane of the pelvis output by Xsens; Body pressure distribution P: Extracting the coordinates of the center of gravity of the seat cushion pressure ( x c , y c (normalized to [0,1]) mean pressure P mean Peak pressure P max ; Steering wheel grip force F (N): The sum of the outputs of the left and right hand grip force sensors, retaining only the steady-state segment lasting >2 seconds; Spinal health and unhealth: Defining health and unhealth (including disease categories).

[0052] (2) Simulation was performed based on the key characteristic parameters of the subjects to obtain the spinal health index and muscle activation of the subjects.

[0053] In practice, parameters such as H, W, and joint length of the subjects are input into a biomechanical simulation model (e.g., Anybody Modeling System), and the joint angle sequence collected by Xsens is used as the motion driver. The simulation outputs the subjects' spinal health indicators and muscle activation: 1) L4 / L5 intervertebral disc axial force (F_axial, unit: N); 2) transverse shear force (F_shear, unit: N); 3) activation of multifidus and erector spinae muscles (%MVC), and electromyography (SEMG) is used for verification.

[0054] (3) A driver spinal health database was constructed based on the key characteristic parameters, spinal health indicators and muscle activation of the subjects, and the weight coefficients of the key characteristic parameters were determined.

[0055] In practical implementation, the collected raw data (height H, weight W, body pressure distribution data P, trunk lean angle A, steering wheel grip force F) are used as input features, and the simulated spinal health indicators (such as L4 / L5 intervertebral disc stress) and spinal disease types are used as output labels. Both are stored in a driver spinal health database to create a structured database. The fields included in each record are shown in Table 1. Database size requirement: ≥2000 valid samples, covering mainstream ergonomic scenarios.

[0056] Table 1. Fields contained in the database

[0057] In one implementation, a dual analysis of Spearman correlation coefficient and random forest feature importance is used to determine the contribution of each key feature parameter to the force on L4 / L5, thus obtaining the weight coefficients of the key feature parameters. For example: trunk forward tilt angle (0.72) > body pressure center of gravity (0.65) > weight (0.48) > grip strength (0.31) > height (0.19). The weight coefficients are used to guide the subsequent matching priority.

[0058] In practical implementation, using Spearman correlation coefficient for feature importance analysis includes: using key feature parameters as input features. X i The forces acting on L4 / L5 are used as the target variables. Y , each feature column X i and target variable Y The numerical values ​​in the table are replaced with their rankings in their respective columns. For example, if a driver's torso lean angle is the 5th smallest among all samples, its value becomes 5. For each feature... X i rank sequence and Y The Pearson correlation coefficient is calculated from the rank sequence, which yields the Spearman correlation coefficient. ρ i ; calculate each ρ i The P-value is calculated. If P > 0.05, the correlation is considered statistically insignificant, and the initial contribution of the feature is recorded as 0 or very low; the absolute value is used as the contribution measure to generate the first set of weight coefficients.

[0059] Feature importance analysis using random forests involves: constructing a random forest regressor, taking key features as input and L4 / L5 forces as output, and setting hyperparameters such as the number of decision trees (usually 100-500), maximum depth, and minimum number of sample splits; after training the model, randomly shuffling a feature... X i Data order (disrupting its relationship with) Y (The relationship between the two features is considered), keeping other features unchanged, re-evaluate the model's error on the test set, and calculate the importance score = (shuffled error - original error). The greater the increase in error, the more important the feature is; normalize the permutation importance scores of all features to the [0,1] interval so that the sum is 1, and generate the second set of weight coefficients.

[0060] Furthermore, the first set of weight coefficients and the second set of weight coefficients are aligned and normalized. Based on the normalized weight coefficients, a linear weighted combination is used to calculate the comprehensive contribution score. The comprehensive contribution score is then normalized again to obtain the final weight coefficients of the key feature parameters.

[0061] To facilitate understanding, this invention also provides a process for constructing a driver spinal health database. First, a driving simulation platform equipped with an adjustable seat (with a travel encoder), a simulated steering wheel (integrated with a grip force sensor), and a display is set up in a temperature- and humidity-controlled laboratory. TekscanConformat pressure-sensing pads are pre-embedded in the seat cushion and backrest. Base stations required for the motion capture system are deployed around the laboratory.

[0062] Furthermore, see Figure 2 The flowchart shown illustrates an experimental data acquisition process. For standard subjects (e.g., male, 30 years old, 175cm tall, 70kg, no history of spinal disease), height, weight, and key joint dimensions such as shoulder width, arm length, and leg length are measured and recorded. Subjects wear Xsens MVN Awinda motion capture suits and complete T-pose and dynamic walking calibrations according to standard procedures. The experimenter adjusts the seat to a preset position (e.g., backrest angle 19°, seat height at the midpoint, fore-aft position moderate). Subjects are instructed to sit in the most natural and comfortable position, lightly gripping the steering wheel with both hands, looking at the monitor in front, and maintaining this posture for 4 minutes. During this time, the system simultaneously records: 1) angle data of all joints throughout the body (especially the T8 thoracic vertebrae and pelvic segment) output by Xsens; 2) pressure distribution data of the seat cushion and backrest output by the pressure sensor pad; and 3) grip force data output by the steering wheel grip force sensor.

[0063] The collected static anthropometric parameters (height, weight, joint length) of the subject were imported into the biomechanical simulation model. Using the Mannequin.any model template, a human body model with the same geometric dimensions as the subject was generated. The Xsens joint angle data (time series) collected from the subject in the first seating posture was used as the kinematic driver input to the personalized model in the biomechanical simulation model. The model's hands were coupled to the steering wheel grip points, and the pelvis was coupled to the seat reference frame.

[0064] Static inverse dynamics simulation was run. Based on the input motion posture, external load (estimated by grip force data for hand force), and the mass properties of the human model itself, Anybody software calculated the forces (axial force and shear force) on each segment of the spine (especially the L4 / L5 intervertebral disc) and the activation level of related core muscle groups (such as multifidus and erector spinae) while maintaining the posture (expressed as a percentage of maximum voluntary contractile force, %MVC), and used electromyography (SEMG) for verification.

[0065] After the simulation is completed, the axial force (e.g., 1520N), shear force (e.g., 210N), and average muscle activation (e.g., 15%MVC) of the L4 / L5 intervertebral disc are exported as the output labels for this record.

[0066] Input features are extracted from the raw data: height (175.0), weight (70.0), trunk lean angle (calculated using T8 and pelvic angle, e.g., 22.5°), seat pressure center of gravity normalized coordinates (x=0.52, y=0.61), and total steady-state grip force (e.g., 35N). These input features, along with the simulation-derived output labels, subject ID, data collection timestamp, and disease type (healthy), are combined according to a predefined database structure to generate a complete record, which is then stored in the driver spinal health database.

[0067] Data collected from the same subject in nine other seating positions was repeated to generate nine new records. This entire process was repeated for at least 199 other subjects to collect data and conduct simulations, ultimately building a database containing no fewer than 2,000 valid records.

[0068] After the database was built, Spearman correlation analysis and random forest regression training (with L4 / L5 intervertebral disc axial force as the prediction target) were performed on all samples in the database using Python's scikit-learn library. The analysis yielded the importance ranking and quantified weights of each input feature, for example: =0.72、 =0.65、 =0.48、 =0.31、 =0.19. These weights will be fixed for fine-tuning and matching at the vehicle end.

[0069] This invention also provides a flowchart for spinal health assessment, see [link / reference]. Figure 3 As shown, the main processes include the following: (1) System startup and driver identification: The driver (taking the same driver in the above embodiment as an example) gets into the vehicle and the vehicle starts. The system confirms the driver's identity through facial recognition or key association and loads the corresponding personal preferences according to actual needs.

[0070] (2) Multi-source data perception: The weighing sensor at the bottom of the seat measures an estimated weight of W=71kg. Based on the seat motor's adjusted memory position and combined with the built-in human proportion model, the height H=176cm is estimated. The depth camera at the front of the roof captures the driver's upper body contour in real time, and through a pre-trained neural network model, the real-time torso forward tilt angle A=20° is estimated. The pressure sensor pad inside the seat collects data in real time, and after processing, the normalized coordinates of the current seat cushion pressure center of gravity are obtained ( x c =0.50, y c =0.60). The steering wheel grip force sensor measured the current total grip force of both hands as F=30N.

[0071] (3) Data preprocessing: The on-board processor performs Kalman filtering on the above raw sensing data to smooth the noise caused by vehicle vibration and driver micro-movement, forming a stable real-time feature vector: [176,71,20,0.50,0.60,30].

[0072] (4) Intelligent database matching: First-level filtering (coarse screening): The matching engine uses the real-time torso tilt angle of 20° as the primary key and queries the `trunk_angle_deg` field, which has a B+ tree index in the database, to quickly find all records with angles in the range of [15°, 25°]. Further filtering conditions of height [171, 181] cm and weight [66, 76] kg are added to filter approximately 120 candidate records from 2000 records.

[0073] Secondary matching (fine screening): For the above 120 candidate records, use the feature weights determined in the previous embodiment to construct a weighted feature vector, and calculate the Mahalanobis distance between the real-time feature vector and the feature vector of each candidate record. Select the K=5 records with the smallest Mahalanobis distance as the most similar neighbors.

[0074] (5) Health assessment output: Read the output labels stored in the 5 most similar neighbor records, namely the axial force values ​​of the L4 / L5 intervertebral disc, assumed to be: [1480N,1500N,1510N,1490N,1520N], and the corresponding Mahalanobis distance is [0.1,0.15,0.2,0.12,0.25].

[0075] Based on the weighted average formula (setting ε=0.001), the estimated axial force of the L4 / L5 intervertebral disc in the current driver is calculated. F end ≈1496N.

[0076] The system's preset scoring rules are as follows: for example, axial force <1000N scores 90-100 (Excellent), 1000-1500N scores 70-89 (Good), 1500-2000N scores 50-69 (Medium), and >2000N scores <50 (Attention). F end ≈1496N, the real-time spinal load score obtained by mapping is "75 points, good".

[0077] Simultaneously, by integrating muscle activation levels and potential historical suggestions from five neighboring records, a prompt is generated: "Current posture is good, core muscle load is normal. Please adjust your sitting posture slightly every hour." These scores, load values, and text prompts are then processed... Figure 4 The HMI interface shown is displayed on the vehicle's infotainment screen for the driver's reference.

[0078] The driver spinal health monitoring method provided in this embodiment of the invention has the following technical effects: (1) By integrating multi-source heterogeneous sensors such as in-vehicle visual posture recognition, seat body pressure distribution array, steering wheel grip force and driver weight measurement, the problem of single data dimension is improved. It can comprehensively and synchronously capture multi-dimensional key biomechanical signals such as body posture, pressure distribution and upper limb load that determine the force on the spine, providing a solid data foundation for accurate assessment.

[0079] (2) The model is dynamically bound to the individual, realizing personalized assessment from a universal model to a one-person-one-model approach: The driver's actual height, weight, joint dimensions, and real-time collected kinematic posture sequences are input into professional biomechanical simulation software to drive the generation of spinal load (such as intervertebral disc stress and muscle activation) labels that strictly correspond to the current individual, and electromyography (SEMG) is used for verification. This solves the drawback of traditional simulation models relying on standard human bodies and being detached from real individuals, making the assessment results highly personalized and accurate.

[0080] (3) Achieving a balance between laboratory-level accuracy and in-vehicle real-time performance through a two-level matching algorithm: Facing the challenge of real-time matching in high-dimensional, large-scale (≥2000 records) databases, a two-level matching algorithm is proposed, combining fast coarse screening based on B+ tree indexing with precise retrieval based on weighted Mahalanobis distance. This algorithm can match the biomechanical model record most similar to the current driver's state from massive amounts of data within 100 milliseconds, thereby seamlessly and in real-time bringing laboratory simulation accuracy into the in-vehicle environment and solving the computational efficiency bottleneck.

[0081] (4) The assessment results are quantified and predictable, achieving a leap from passive alarm to proactive health management: The system output is no longer a simple threshold alarm, but a quantified spinal load score (0-100 points), specific intervertebral disc stress values ​​(unit: N), and muscle activation level (%MVC). Combined with the matching results, specific and actionable health recommendations can also be generated (such as "It is recommended to increase the backrest tilt angle by 3°"). This enables drivers to scientifically and intuitively understand their own spinal load status and risk trends, thereby proactively adjusting their driving posture and achieving true predictive health management.

[0082] Regarding the driver spinal health monitoring method provided in the foregoing embodiments, the present invention provides a driver spinal health monitoring system, see [link to documentation]. Figure 5 The diagram shown illustrates a method for monitoring driver spinal health, indicating that the method mainly includes the following components: The data acquisition module 501 is used to collect the driver's physiological data through on-board multi-source sensors; The feature extraction module 502 is used to preprocess physiological data and extract features to obtain the driver's feature parameters; The retrieval module 503 is used to retrieve the driver's spinal health indicators based on feature parameters and a pre-built driver spinal health database.

[0083] The driver spinal health monitoring system provided in this embodiment of the invention, by integrating multiple in-vehicle sensors, can comprehensively and synchronously capture multi-dimensional key physiological data such as body posture, pressure distribution, and upper limb load that determine spinal force, thus improving the problem of single data dimension in the prior art; by retrieving the driver's spinal health index through a pre-built driver spinal health database, it can provide drivers with personalized quantitative assessment of spinal health, thereby improving the accuracy of driver spinal health monitoring.

[0084] In one implementation, the feature parameters include at least: height, weight, torso lean angle, center of gravity, and steering wheel grip strength; the retrieval module 503 is specifically used to: perform a first-level search in a pre-constructed driver spinal health database based on the torso lean angle to obtain a first candidate sample; calculate a weighted feature vector of the feature parameters based on a pre-set weight coefficient, and perform a second-level search in the first candidate sample based on the weighted feature vector to obtain a preset number of second candidate samples; and calculate the driver's spinal health index based on the spinal health index of the second candidate samples.

[0085] In one embodiment, the retrieval module 503 is specifically used to: calculate the difference between the torso lean angle and the torso lean angle of each sample in the driver's spinal health database; determine the samples whose absolute value of the difference is less than or equal to a preset value as candidate samples; and filter the candidate samples based on height and weight to obtain the first candidate sample.

[0086] In one implementation, the retrieval module 503 is specifically used to: calculate the similarity between the weighted feature vector and the weighted feature vector of each sample in the first candidate sample; and select a preset number of first candidate samples as second candidate samples based on the similarity.

[0087] In one implementation, the retrieval module 503 is specifically used to: perform a weighted average of the spinal health indicators of the second candidate sample based on similarity to obtain the spinal health indicators of the driver.

[0088] In one embodiment, the system further includes a scoring module, configured to: determine the driver's spinal health score based on the driver's spinal health indicators and preset scoring rules; generate the driver's core muscle load status and prompt information based on the muscle activation of a second candidate sample; and display the driver's spinal health score, core muscle load status, and prompt information on the vehicle's infotainment screen.

[0089] In one embodiment, the system further includes a database construction module, used to: collect physiological data of test subjects and extract key feature parameters from the physiological data; wherein the key feature parameters include at least: height, weight, joint length, trunk lean angle, center of gravity, steering wheel grip strength, and spinal health status; perform simulation based on the key feature parameters of the test subjects to obtain the spinal health indicators and muscle activation of the test subjects; construct a driver spinal health database based on the key feature parameters, spinal health indicators, and muscle activation of the test subjects, and determine the weight coefficients of the key feature parameters.

[0090] It should be noted that the system provided in this embodiment of the invention has the same implementation principle and technical effects as the aforementioned method embodiment. For the sake of brevity, any parts not mentioned in the system embodiment can be referred to the corresponding content in the aforementioned method embodiment. The specific numerical values ​​provided in the implementation of this invention are merely exemplary and are not intended to limit the scope of the invention.

[0091] This invention also provides an intelligent cockpit for automobiles. Specifically, the intelligent cockpit includes a processor and a storage device; the storage device stores a computer program, which, when run by the processor, executes the method described in any of the above embodiments.

[0092] Figure 6This is a schematic diagram of the structure of an intelligent cockpit for automobiles provided in an embodiment of the present invention. The intelligent cockpit 100 includes: a processor 60, a memory 61, a bus 62, and a communication interface 63. The processor 60, the communication interface 63, and the memory 61 are connected through the bus 62. The processor 60 is used to execute executable modules, such as computer programs, stored in the memory 61.

[0093] The memory 61 may include high-speed random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 63 (which can be wired or wireless), such as the Internet, wide area network, local area network, metropolitan area network, etc.

[0094] Bus 62 can be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 6 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0095] The memory 61 is used to store programs. After receiving an execution instruction, the processor 60 executes the program. The method executed by the device for defining the flow process disclosed in any of the foregoing embodiments of the present invention can be applied to the processor 60 or implemented by the processor 60.

[0096] Processor 60 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of processor 60 or by instructions in software form. Processor 60 can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this invention. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory 61. Processor 60 reads the information in memory 61 and, in conjunction with its hardware, completes the steps of the above method.

[0097] The computer program product of the readable storage medium provided in the embodiments of the present invention includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods described in the foregoing method embodiments. For specific implementation, please refer to the foregoing method embodiments, which will not be repeated here.

[0098] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0099] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for monitoring the spinal health of drivers, characterized in that, include: Physiological data of the driver is collected through onboard multi-source sensors; The physiological data is preprocessed and features are extracted to obtain the driver's feature parameters; Based on the aforementioned feature parameters and a pre-built database of driver spinal health, the driver's spinal health indicators are obtained.

2. The method according to claim 1, characterized in that, The characteristic parameters include at least: height, weight, torso lean angle, center of gravity, and steering wheel grip strength; Based on the aforementioned feature parameters and a pre-built driver spinal health database, the driver's spinal health indicators are obtained, including: Based on the trunk forward tilt angle, a first-level search is performed in a pre-constructed driver spinal health database to obtain the first candidate sample; The weighted feature vector of the feature parameters is calculated based on the pre-set weight coefficients, and a secondary search is performed on the first candidate sample based on the weighted feature vector to obtain a preset number of second candidate samples. The driver's spinal health index was calculated based on the spinal health index of the second candidate sample.

3. The method according to claim 2, characterized in that, Based on the aforementioned forward tilt angle, a primary search is performed in a pre-constructed driver spinal health database to obtain the first candidate samples, including: Calculate the difference between the trunk forward tilt angle and the trunk forward tilt angle of each sample in the driver spinal health database; Samples whose absolute value of the difference is less than or equal to a preset value are identified as candidate samples; The candidate samples are filtered based on the height and weight to obtain the first candidate sample.

4. The method according to claim 2, characterized in that, Based on the weighted feature vector, a secondary search is performed on the first candidate samples to obtain a predetermined number of second candidate samples, including: Calculate the similarity between the weighted feature vector and the weighted feature vector of each sample in the first candidate sample; Based on the similarity, a preset number of first candidate samples are selected as second candidate samples.

5. The method according to claim 4, characterized in that, The driver's spinal health indicators are calculated based on the spinal health indicators of the second candidate sample, including: Based on the similarity, a weighted average of the spinal health indicators of the second candidate sample is performed to obtain the spinal health indicators of the driver.

6. The method according to claim 2, characterized in that, After retrieving the driver's spinal health indicators based on the aforementioned feature parameters and a pre-built driver spinal health database, the process further includes: Based on the driver's spinal health indicators and preset scoring rules, the driver's spinal health score is determined. Based on the muscle activation level of the second candidate sample, the driver's core muscle group load status and prompt information are generated; The driver's spinal health score, core muscle load status, and prompts are displayed on the vehicle's infotainment screen.

7. The method according to claim 1, characterized in that, Before collecting the driver's physiological data using onboard multi-source sensors, the following steps are also included: Physiological data of the subjects were collected, and key feature parameters were extracted from the physiological data; wherein, the key feature parameters include at least: height, weight, joint length, trunk forward tilt angle, center of gravity of pressure, steering wheel grip strength, and spinal health status. Simulations were performed based on the key characteristic parameters of the subjects to obtain their spinal health indicators and muscle activation levels. A driver spinal health database is constructed based on the key characteristic parameters of the subjects, the spinal health indicators, and the muscle activation, and the weight coefficients of the key characteristic parameters are determined.

8. A driver spinal health monitoring system, characterized in that, include: The data acquisition module is used to collect the driver's physiological data through onboard multi-source sensors; The feature extraction module is used to preprocess and extract features from the physiological data to obtain the driver's feature parameters; The retrieval module is used to retrieve the driver's spinal health indicators based on the feature parameters and a pre-built driver spinal health database.

9. An intelligent cockpit for automobiles, characterized in that, The method includes a processor and a memory, the memory storing computer-executable instructions executable by the processor, the processor executing the computer-executable instructions to implement the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program thereon, characterized in that, The computer program is executed by the processor to perform the steps of the method described in any one of claims 1 to 7.

Images