Distraction driving intervention method and system based on subthreshold vibration stimulation

Abstract

The invention relates to the technical field of driving safety, and discloses a distracted driving intervention method and system based on subthreshold vibration stimulation, and the method comprises the steps: obtaining eye movement data of a driver in a driving process; preprocessing the eye movement data, and extracting eye movement features; according to the trained SVM model, classifying the eye movement features to judge whether the driver is distracted or not; if it is determined that the driver is distracted, subthreshold vibration stimulation is initiated to operate the vibrator at a set frequency and set intensity. Eye movement data can be collected in the driving process, the distraction driving state of a driver is judged, and distraction early warning is conducted on the driver in a subthreshold vibration stimulation mode. Compared with the traditional voice and video early warning, the subthreshold vibration stimulation adopted by the invention can enable the driver to recover the attention while not increasing the cognitive load of the driver.

Description

Technical Field

[0001] The present invention relates to the technical field of driving safety, and in particular to a distracted driving intervention method and system based on subthreshold vibration stimulation. Background Art

[0002] Distracted driving is one of the major challenges facing modern traffic safety and has become an important factor leading to traffic accidents and related casualties. Distracted driving refers to the driver's distraction due to non-driving tasks (such as using mobile phones, adjusting on-board equipment, eating, etc.) during driving, and he cannot focus on driving operations, thereby increasing the risk of accidents. Studies have shown that distracted driving has a negative impact on the driver's cognition, vision and motor control, such as prolonging reaction time, reducing vehicle control ability and judgment, etc. With the popularization of smartphones and other electronic devices, the frequency of distracted driving has increased significantly, especially among young drivers. This phenomenon not only poses a threat to the driver himself, but also has a serious impact on the safety of other road users. Therefore, how to effectively prevent and reduce distracted driving behavior has become an important issue in road traffic safety research and policy making.

[0003] Eye movement data is widely used in distraction detection in driving, flying, medical and other fields. It evaluates the individual's attention level by monitoring features such as eye fixation time, eye saccade frequency, and pupil changes. When distracted, these eye movement features often show abnormal changes, such as shortened fixation time and frequent eye saccades. With the development of eye tracking technology, real-time and accurate distraction detection has become possible.

[0004] Subthreshold vibration stimulation is a new warning method used in the fields of driving assistance and attention guidance. The so-called subthreshold stimulation refers to stimulation whose intensity is lower than the perception threshold and cannot be obviously perceived by people, but it can still have an impact on the individual's behavior. Subthreshold vibration stimulation can enhance the driver's alertness to potential dangers without interfering with the driver's explicit attention and cognitive resources by applying vibration signals within a range below the driver's attention perception level. Compared with traditional visual or auditory warnings, subthreshold vibration stimulation does not increase the driver's cognitive burden, and can reduce the explicit distraction of the driver's attention. It is an effective and covert auxiliary warning method with great application potential in complex driving scenarios.

[0005] Patent No. CN112998719 proposes a driver status monitoring device that collects the driver's facial expressions and EEG signals during driving, conducts real-time analysis and judgment of abnormal conditions such as drunkenness, fatigue, distraction, etc., and provides screen display and sound reminders. When the driver's attention is focused on areas other than the front, the visual warning of this driver status monitoring device may not work effectively, thereby posing a risk of missing key warning information. In addition, the diverse auditory interference in the environment (such as in-car conversations or road noise) may burden the driver's auditory system, making it difficult for the sound warning to be perceived and responded to effectively in a timely manner. When the auditory system is under increased load, it is easy for the warning information to be masked or ignored, thereby reducing the reliability and effectiveness of the sound warning.

[0006] Patent No. CN117719532 proposes a multi-channel vehicle seat vibration feedback system, which generates multi-channel drive signals for different interaction scenarios between the car and the driver and passengers, and controls the vibration and sound of the multi-functional car seat by controlling the output channel separately, so that the signal feedback to the driver and passengers is more accurate and rich, improving the user experience. Although the driver can obtain early warning information through the vibration signal, the driver still needs to analyze and interpret the vibration signal, which increases the cognitive load.

[0007] Therefore, how to help drivers regain their attention without increasing cognitive load is a technical problem that urgently needs to be solved. Summary of the invention

[0008] The purpose of the present invention is to provide a distracted driving intervention method and system based on subthreshold vibration stimulation, which determines whether distraction occurs by collecting the driver's eye movement data during driving, and provides an attention recovery method that does not increase cognitive load when distraction occurs.

[0009] In order to achieve the above purpose, the technical solutions adopted are as follows: In a first aspect, the present invention provides a distracted driving intervention method based on subthreshold vibration stimulation, the method comprising: Obtain the driver's eye movement data during driving; Preprocessing the eye movement data and extracting eye movement features; Classifying the eye movement features according to the trained SVM model to determine whether the driver is distracted; In the case where it is determined that the driver is distracted, the subthreshold vibration stimulation is activated so that the vibrator operates at a set frequency and a set intensity.

[0010] Furthermore, the eye movement data is preprocessed and eye movement features are extracted, including: Detecting and removing invalid data items in the eye movement data; wherein the invalid data items include null values ​​and data points whose coordinates exceed a set range; and based on a preset signal quality threshold, filtering the eye movement data, removing data points whose signal quality is lower than the preset signal quality threshold, and obtaining first eye movement data; Performing low-pass filtering on the first eye movement data to remove high-frequency noise, and adjusting the sampling frequency of the data to remove redundant high-frequency data, to obtain second eye movement data; When the eye movement coordinates in the second eye movement data exceed a set standard deviation threshold, determining that the eye movement coordinates in the second eye movement data are abnormal values, and correcting the abnormal values ​​based on linear interpolation to obtain third eye movement data; wherein the third eye movement data includes eye movement velocity and acceleration data; Based on the set speed threshold, if the eye movement speed is lower than the set speed threshold, it is identified as the fixation phase of the eye movement event; if the eye movement speed is not lower than the set speed threshold, it is identified as the scanning phase of the eye movement event; the duration of the fixation phase is determined according to the time period of the eye movement coordinate change, and the scanning speed is calculated from the rate of change of the eye movement trajectory; the duration of the fixation phase and the scanning speed are used as eye movement features.

[0011] Furthermore, the trained SVM model is trained by the following method: Extracting feature data from the eye movement features; wherein the feature data includes blinking frequency, fixation duration of the AOI area, and pupil size; Based on the feature data, divide the data into a training set and a test set in proportion; Select radial basis function as the kernel function of the SVM model; Adjust the penalty parameters and kernel parameters of the SVM model to optimize the performance of the classifier and obtain the trained SVM model; The trained SVM model is tested using the test set, and if the test passes, the currently trained SVM model is used as the trained SVM model.

[0012] Furthermore, when adjusting the penalty parameters and kernel parameters of the SVM model, cross-validation is used to find the optimal parameter combination, and the SVM model is configured with the optimal parameter combination so that the SVM model can recognize the distraction feature patterns in the eye movement data.

[0013] Furthermore, the trained SVM model is tested using a test set, including: The test set is input into the trained SVM model, and the trained SVM model outputs a distraction state, which includes distraction and non-distraction. The distraction state is determined by a threshold method. If the classification output value is greater than the set threshold, it is determined to be distraction; if the classification output value is not greater than the set threshold, it is determined to be non-distraction.

[0014] Further, the vibrator includes a steering wheel vibrator and / or a seat vibrator; wherein, the steering wheel vibrator includes a first LRA vibrator, the seat vibrator includes a second LRA vibrator, the first LRA vibrator is embedded and installed in the left and right handle areas of the steering wheel, and the second LRA vibrator is embedded and installed in the back of the driver's seat corresponding to the human lumbar area and / or the positions on both sides of the driver's seat corresponding to the human hip area.

[0015] Further, the first LRA vibrator is set at a frequency of 20-50 Hz to deliver vibration stimulation at a subconscious level, thereby reminding the driver to concentrate; The setting frequency of the second LRA vibrator is 75-100 Hz to improve the driver's alertness during long-term driving or fatigue.

[0016] Further, the set frequency of the first LRA vibrator is 25 Hz to match the sensitive frequency band of the Messner corpuscles in the hand; The setting frequency of the second LRA vibrator is 85 Hz to match the sensitive frequency band of the Pacinian corpuscles in the lower back.

[0017] Further, the set intensities of the first LRA vibrator and the second LRA vibrator are both set to 0.05-0.1g, where g represents gravity acceleration; the set intensities of the first LRA vibrator and the second LRA vibrator are adjusted according to individual differences and feedback from drivers, and / or feedback and response data from drivers.

[0018] In a second aspect, the present invention provides a distracted driving intervention system based on subthreshold vibration stimulation, the system comprising: A data acquisition module is configured to obtain eye movement data of the driver during driving; A data preprocessing module, configured to preprocess the eye movement data and extract eye movement features; A distraction prediction module is configured to classify the eye movement features according to a trained SVM model to determine whether the driver is distracted; The subthreshold vibration stimulation module is configured to start the subthreshold vibration stimulation when it is determined that the driver is distracted, so that the vibrator works at a set frequency and a set intensity.

[0019] The beneficial effects of the present invention are: The present invention determines the driver's distracted state by collecting eye movement data during vehicle driving. In view of the risk of increasing the driver's cognitive load in the existing warning mode, a subthreshold vibration stimulation attention recovery mode is proposed, which allows the driver to recover from the distracted state without increasing the cognitive load, and warns of potential dangerous situations. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A structural diagram of a distracted driving intervention device based on subthreshold vibration stimulation according to an embodiment of the present invention is shown.

[0021] Figure 2 A flow chart of a distracted driving intervention method based on subthreshold vibration stimulation according to an embodiment of the present invention is shown.

[0022] Figure 3 A flow chart of preprocessing eye movement data according to an embodiment of the present invention is shown.

[0023] Figure 4 A training flow chart of an SVM model according to an embodiment of the present invention is shown.

[0024] Figure 5 A structural diagram of a steering wheel vibrator according to an embodiment of the present invention is shown.

[0025] Figure 6 A structural diagram of a driver's seat according to an embodiment of the present invention is shown.

[0026] Figure 7 A structural diagram of a distracted driving intervention system based on subthreshold vibration stimulation according to an embodiment of the present invention is shown.

[0027] Description of reference numerals: 101-wearable eye tracker; 102-data processing chip; 103-vibrator; 51-first LRA vibrator; 52-steering wheel; 521-left handle area; 522-right handle area; 611-the back of the driver's seat corresponds to the position of the human lumbar region; 612-the positions of both sides of the driver's seat corresponding to the human hips; 62-driver's seat; 701-data acquisition module; 702-data preprocessing module; 703-distraction prediction module; 704-subthreshold vibration stimulation module DETAILED DESCRIPTION

[0028] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0029] The specific implementation of the present invention is further described in detail below in conjunction with the drawings and examples.

[0030] Figure 1 FIG. 1 shows a structural diagram of a distracted driving intervention device based on subthreshold vibration stimulation according to an embodiment of the present invention, see Figure 1 The distracted driving intervention device based on subthreshold vibration stimulation is used to implement distracted driving intervention, and includes a wearable eye tracker 101, a data processing chip 102 and a vibrator 103; wherein the wearable eye tracker 101 is worn by a driver to collect eye movement data of the driver during driving, the wearable eye tracker 101 is connected to the data processing chip 102 to feed the collected eye movement data of the driver during driving to the data processing chip 102, the data processing chip 102 is connected to the vibrator 103, the data processing chip 102 determines whether the driver is distracted based on the eye movement data, and when it is determined that the driver is distracted, controls the vibrator 103 to work at a set frequency and a set intensity to remind the driver.

[0031] The data processing chip 102 is configured to implement a distracted driving intervention method based on subthreshold vibration stimulation, that is, the data processing chip 102 is configured to obtain the driver's eye movement data during driving; pre-process the eye movement data and extract eye movement features; classify the eye movement features according to the trained SVM model to determine whether the driver is distracted; and when it is determined that the driver is distracted, start the subthreshold vibration stimulation to make the vibrator work at a set frequency and a set intensity. The present invention can collect eye movement data during driving, identify the driver's distracted driving state, and use subthreshold vibration stimulation to give the driver a distraction warning.

[0032] Taking a driver driving a car as an example, the data processing chip 102 can be installed on the center control panel of the car, or the data processing chip 102 can be integrated on the wearable eye tracker 101. When the data processing chip 102 can be installed on the center control panel of the car, the data processing chip 102 establishes a signal connection with the wearable eye tracker 101 through a wireless transmission module. When the data processing chip 102 can be integrated on the wearable eye tracker 101, the data processing chip 102 directly uses a wire to establish a signal connection with the wearable eye tracker 101; the vibrator 103 is set to be multiple, and is respectively embedded in the steering wheel and the driver's seat. The connection method between the vibrator 103 and the data processing chip 102 can be wireless or wired, wherein the wireless connection is to establish a connection between the vibrator 103 and the data processing chip 102 through a wireless transmission module, and the wireless transmission module includes but is not limited to a WiFi module, a Bluetooth module, a 4G / 5G module, etc.

[0033] It should be noted that the above application in automobiles is only an example and is not a limitation of the present invention. In other embodiments, the present invention can be applied to other means of transportation, such as ships, airplanes, trains, etc.

[0034] It should be noted that the wearable eye tracker 101, the data processing chip 102 and the vibrator 103 are all existing product components, and their specific structures are not expanded here in this embodiment. The present invention has two core points. The first is the process of identifying the distraction state based on eye movement data, and the second is to use tactile perception to remind the driver based on the distraction state after identification, and use different frequencies and intensities for different tactile parts (hands, back, etc.). Using specific frequencies and intensities for specific parts can give the driver a way to restore attention without increasing cognitive load when distraction occurs. This method can effectively reduce the driver's attention distraction and does not increase his cognitive burden, thereby significantly improving the driver's attention resource management efficiency.

[0035] Figure 2 FIG. 4 is a flowchart showing a distracted driving intervention method based on subthreshold vibration stimulation according to an embodiment of the present invention. Figure 2 The embodiment of the present invention provides a distracted driving intervention method based on subthreshold vibration stimulation. The distracted driving intervention method based on subthreshold vibration stimulation can be based on the following Figure 1 The device shown is implemented, for example, by configuring the method in a data processing chip 102 as a computer program product, so that eye movement data can be collected during driving, the driver's distracted driving state can be judged, and the driver can be warned of distraction by using subthreshold vibration stimulation. The distracted driving intervention method based on subthreshold vibration stimulation includes steps S10 to S40, which are described in detail as follows.

[0036] S10: Acquire the driver's eye movement data during driving.

[0037] In this embodiment, the driver's eye movement data during driving can be collected by the wearable eye tracker 101. For example, in a real driving environment, the driver is asked to wear the wearable eye tracker 101 to record the eye movement data during driving. In driver distraction detection, eye movement features are widely used to judge the driver's attention state, mainly because human eye movement behavior directly reflects the changes in their visual attention and cognitive load.

[0038] S20, preprocessing the eye movement data and extracting eye movement features.

[0039] In some embodiments, Figure 3 FIG. 4 shows a flow chart of preprocessing eye movement data according to an embodiment of the present invention. Figure 3 , step S20 includes steps S201-S204, which are described in detail as follows.

[0040] S201. Detect and remove invalid data items in the eye movement data; wherein the invalid data items include null values ​​and data points whose coordinates exceed a set range; and based on a preset signal quality threshold, filter the eye movement data to remove data points whose signal quality is lower than the preset signal quality threshold, and obtain first eye movement data.

[0041] In this embodiment, the purpose of step S201 is to clean up data and remove invalid data. Specifically, invalid data items in the eye movement data are detected and removed, including null values ​​(NaN values) and points whose coordinates exceed the valid range. For example, when the eye movement coordinates exceed the screen boundary or are invalid values, they are regarded as invalid data and removed. According to the signal quality index provided by the eye movement acquisition device (such as the wearable eye tracker 101), a signal quality threshold (such as less than 0.5) is set to filter the data and remove data points with low signal quality.

[0042] S202 , low-pass filtering the first eye movement data to remove high-frequency noise, and adjusting the sampling frequency of the data to remove redundant high-frequency data, to obtain second eye movement data.

[0043] In this embodiment, step S202 is a step of noise removal and signal filtering. Specifically, the first eye movement data is low-pass filtered to remove high-frequency noise, such as using a Butterworth filter to smooth the eye movement data. The sampling frequency of the data is further adjusted to remove redundant high-frequency data, and only representative eye movement data is retained as the second eye movement data.

[0044] S203. When the eye movement coordinates in the second eye movement data exceed the set standard deviation threshold, determine that the eye movement coordinates in the second eye movement data are abnormal values, and correct the abnormal values ​​based on linear interpolation to obtain third eye movement data; wherein the third eye movement data includes eye movement velocity and acceleration data.

[0045] In this embodiment, step S203 is a step of outlier detection and correction. Specifically, Z-Score is used to detect outliers. When the change of eye movement coordinates exceeds the set standard deviation threshold, the data point is considered to be an outlier. For the detected outliers, an interpolation method is used for correction. The interpolation method can use linear interpolation to ensure data continuity and avoid data loss.

[0046] S204. Based on the set speed threshold, if the eye movement speed is lower than the set speed threshold, it is identified as the fixation phase of the eye movement event; if the eye movement speed is not lower than the set speed threshold, it is identified as the scanning phase of the eye movement event; the duration of the fixation phase is determined according to the time period of the eye movement coordinate change, and the scanning speed is calculated by the rate of change of the eye movement trajectory; the duration of the fixation phase and the scanning speed are used as eye movement features.

[0047] In this embodiment, step S204 identifies eye movement events by the third eye movement data obtained in step S203, and extracts eye movement features based on the identified eye movement events. Specifically, based on the eye movement velocity and acceleration data, the fixation and saccade phases in the eye movement events are identified. As an example only, the fixation and saccade phases in the eye movement events are identified by the following method: if the eye movement velocity is lower than a predetermined threshold (such as 30° / s), it is identified as fixation; if the eye movement velocity is higher than the threshold, it is saccade.

[0048] After identifying the fixation and saccade phases in an eye movement event, the duration of each fixation phase was calculated, and the velocity data of the saccade phase was extracted. The duration of fixation was determined by the time period of eye movement coordinate change, and the saccade velocity was calculated by the rate of change of the eye movement trajectory.

[0049] S30. Classify the eye movement features according to the trained SVM model to determine whether the driver is distracted.

[0050] In some embodiments, Figure 4 FIG. 4 shows a training flow chart of an SVM model according to an embodiment of the present invention. Figure 4 , step S30 includes steps S301-S305, which are described in detail as follows.

[0051] S301, extracting feature data from eye movement features; wherein the feature data includes blinking frequency, fixation duration of the AOI area, and pupil size.

[0052] S302: Based on the feature data, divide the data into a training set and a test set in proportion.

[0053] Exemplarily, when the feature data is divided into a training set and a test set in proportion, a ratio of 7:3 is used for data segmentation; wherein the training set is used for model training, and the test set is used for model testing.

[0054] S303 , selecting a radial basis function as a kernel function of the SVM model.

[0055] In this embodiment, the radial basis function is suitable for processing nonlinear characteristics and can better separate the driver's distraction state.

[0056] S304: Adjust the penalty parameters and kernel parameters of the SVM model to optimize the performance of the classifier and obtain a trained SVM model.

[0057] In this embodiment, the parameter adjustment method may use cross-validation to find the optimal parameter combination. The SVM model is trained using the training set data so that it can identify the distraction feature pattern in the eye movement data.

[0058] S305: Use the test set to test the trained SVM model. If the test passes, the currently trained SVM model is used as the trained SVM model.

[0059] In some embodiments, the test set is input into the trained SVM model, and the trained SVM model outputs the distraction state, which includes distraction and non-distraction. The distraction state is determined by the threshold method. If the classification output value is greater than the set threshold, it is determined to be distraction; if the classification output value is not greater than the set threshold, it is determined to be non-distraction.

[0060] In this embodiment, the method of judging whether the test passes can be based on the test indicators and the corresponding indicator thresholds. For example, the test indicators can be selected from at least one of the accuracy, precision, recall, F1 value, AUC and ROC curve, confusion matrix, mean square error, mean absolute error and root mean square error and their combinations.

[0061] S40: When it is determined that the driver is distracted, start subthreshold vibration stimulation to make the vibrator work at a set frequency and a set intensity.

[0062] In some embodiments, the vibrator 103 includes a steering wheel vibrator and / or a seat vibrator. In specific implementations, the vibrator 103 may include only one steering wheel vibrator or one seat vibrator, in which case vibration reminders are only provided for the hands or back or hips. The vibrator 103 may also include both a steering wheel vibrator and a seat vibrator, in which case vibration reminders are provided for the hands, back, and hips at the same time.

[0063] Figure 5 2 shows a structural diagram of a steering wheel vibrator according to an embodiment of the present invention. Figure 5 The steering wheel vibrator includes two first LRA vibrators 51 , and the two first LRA vibrators 51 are respectively embedded and installed in a left grip area 521 and a right grip area 522 of a steering wheel 52 .

[0064] It should be noted that the first LRA vibrator 51 is a linear resonant vibrator (Linear Resonant Actuator, LRA), which is small in size, fast in response speed, and has a controllable vibration frequency. The first LRA vibrator 51 is installed in the left grip area 521 and the right grip area 522 of the steering wheel 52 in an embedded manner (i.e., the 9 o'clock and 3 o'clock positions when the steering wheel 52 is straightened). The first LRA vibrator 51 is close to the surface of the steering wheel 52 to ensure that the vibration can be effectively transmitted to the driver's hands. Through the embedded installation method, the first LRA vibrator 51 is fixed in the internal structure of the steering wheel 52 to ensure stability and durability, and does not affect the normal steering function of the steering wheel 52.

[0065] in addition, Figure 5 The shape of the steering wheel 52 shown is only an example, and does not mean that the steering wheel vibrator must be applied to this type of steering wheel.

[0066] The seat vibrator includes a second LRA vibrator, Figure 6 2 shows a structural diagram of a driver's seat according to an embodiment of the present invention. Figure 6 The second LRA vibrator is embedded and installed at a position 611 on the back of the driver's seat 62 corresponding to the human lumbar region and / or at a position 612 on both sides of the driver's seat 62 corresponding to the human hip region.

[0067] In this embodiment, the second LRA vibrator is also an LRA vibrator, which is suitable for the space limitation inside the seat. The second LRA vibrator is embedded in the back of the driver's seat 62 at the position 611 corresponding to the human lumbar region and / or at the positions 612 on both sides corresponding to the human hips, so that the vibration stimulation can be transmitted to the driver's waist and back. Exemplarily, the installation method is: the second LRA vibrator is embedded in the seat filling material, and the original seat fabric is covered on the outside to maintain the comfort and aesthetics of the seat.

[0068] After the distracted driving state is determined, this embodiment uses subthreshold vibration stimulation to warn the driver of distraction, so as to restore the driver's attention without increasing the driver's cognitive load. Among them, the subthreshold vibration stimulation method mainly uses the set frequency and intensity to control the steering wheel vibrator and / or the seat vibrator to vibrate and remind the driver. The selection basis and the mechanism of action of the vibration frequency and intensity of different vibrators in the subthreshold vibration stimulation method will be described in detail below.

[0069] First of all, the physiological basis for the selection of vibration frequency is that the tactile receptors in human skin have different sensitivities to vibrations of different frequencies. Meissner corpuscles are sensitive to low-frequency vibrations of 20-50 Hz and are mainly distributed in the skin of fingers and palms. Pacinian corpuscles are sensitive to high-frequency vibrations of 60-400 Hz and are widely distributed in deep skin and tissues.

[0070] Secondly, the subthreshold stimulation principle: the selected vibration frequency should activate the corresponding receptors but not exceed their perception threshold to avoid causing explicit perception of the driver.

[0071] The vibration frequency is set based on the subthreshold stimulation principle.

[0072] For Steering Wheel Vibrator: Frequency range: set between 20-50 Hz, preferably 25 Hz, to match the sensitive frequency band of the Messner corpuscles in the hand.

[0073] Effect: Delivers vibration stimulation at the subconscious level to remind the driver to concentrate.

[0074] For seat vibrators: Frequency range: Set to 75-100 Hz, preferably 85 Hz, to match the sensitive frequency band of the Pacinian corpuscles in the lower back.

[0075] Effect: Improve the driver's alertness during long-term driving or fatigue.

[0076] The vibration intensity settings of the steering wheel vibrator and the seat vibrator are consistent. For the intensity range: the vibration acceleration is set between 0.05-0.1g to ensure that the stimulation is at a subthreshold level, where g represents the acceleration of gravity. Intensity adjustment: According to the individual differences and feedback of the driver, the intensity is appropriately adjusted to ensure that it does not cause obvious perception and can effectively transmit the warning signal. Personalized adjustment: According to the feedback and response data of the driver, the vibration frequency and intensity are adjusted to establish a personalized warning parameter configuration.

[0077] In some embodiments, the steering wheel vibrator and the seat vibrator may operate in a manner that vibrations may be applied continuously or intermittently to avoid adaptation and maintain stimulation effectiveness.

[0078] The present invention also provides a distracted driving intervention system based on subthreshold vibration stimulation, see Figure 7 The system comprises: The data acquisition module 701 is configured to obtain the driver's eye movement data during driving; A data preprocessing module 702 is configured to preprocess the eye movement data and extract eye movement features; The distraction prediction module 703 is configured to classify the eye movement features according to the trained SVM model to determine whether the driver is distracted; The subthreshold vibration stimulation module 704 is configured to start the subthreshold vibration stimulation when it is determined that the driver is distracted, so as to make the vibrator work at a set frequency and a set intensity.

[0079] It should be noted that the distracted driving intervention system based on subthreshold vibration stimulation belongs to the same technical concept as the previously described method, and has the same technical principles and beneficial effects, so it will not be repeated here.

[0080] The above implementation modes are only used to illustrate the present invention, but not to limit the present invention. Ordinary technicians in the relevant technical field can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, all equivalent technical solutions also belong to the scope of the present invention. The patent protection scope of the present invention should be defined by the claims.

Claims

1. A distracted driving intervention method based on subthreshold vibration stimulation, characterized in that: The method comprises: Obtain the driver's eye movement data during driving; Preprocessing the eye movement data and extracting eye movement features; Classifying the eye movement features according to the trained SVM model to determine whether the driver is distracted; In the case where it is determined that the driver is distracted, the subthreshold vibration stimulation is initiated so that the vibrator operates at a set frequency and a set intensity; Preprocessing the eye movement data and extracting eye movement features includes: Detecting and removing invalid data items in the eye movement data; wherein the invalid data items include null values ​​and data points whose coordinates exceed a set range; and based on a preset signal quality threshold, filtering the eye movement data, removing data points whose signal quality is lower than the preset signal quality threshold, and obtaining first eye movement data; Performing low-pass filtering on the first eye movement data to remove high-frequency noise, and adjusting the sampling frequency of the data to remove redundant high-frequency data, to obtain second eye movement data; When the eye movement coordinates in the second eye movement data exceed a set standard deviation threshold, determining that the eye movement coordinates in the second eye movement data are abnormal values, and correcting the abnormal values ​​based on linear interpolation to obtain third eye movement data; wherein the third eye movement data includes eye movement velocity and acceleration data; Based on the set speed threshold, if the eye movement speed is lower than the set speed threshold, it is identified as the fixation phase of the eye movement event; if the eye movement speed is not lower than the set speed threshold, it is identified as the scanning phase of the eye movement event; the duration of the fixation phase is determined according to the time period of the eye movement coordinate change, and the scanning speed is calculated from the rate of change of the eye movement trajectory; the duration of the fixation phase and the scanning speed are used as eye movement features.

2. The method according to claim 1, characterized in that The trained SVM model is trained by the following method: Extracting feature data from the eye movement features; wherein the feature data includes blinking frequency, fixation duration of the AOI area, and pupil size; Based on the feature data, divide the data into a training set and a test set in proportion; Select radial basis function as the kernel function of the SVM model; Adjust the penalty parameters and kernel parameters of the SVM model to optimize the performance of the classifier and obtain the trained SVM model; The trained SVM model is tested using the test set, and if the test passes, the currently trained SVM model is used as the trained SVM model.

3. The method according to claim 2, characterized in that When adjusting the penalty parameters and kernel parameters of the SVM model, cross-validation is used to find the optimal parameter combination, and the SVM model is configured with the optimal parameter combination so that the SVM model can identify distraction feature patterns in eye movement data.

4. The method according to claim 2, characterized in that The trained SVM model is tested using a test set, including: The test set is input into the trained SVM model, and the trained SVM model outputs a distraction state, which includes distraction and non-distraction. The distraction state is determined by a threshold method. If the classification output value is greater than the set threshold, it is determined to be distraction; if the classification output value is not greater than the set threshold, it is determined to be non-distraction.

5. The method according to claim 4, characterized in that The vibrator includes a steering wheel vibrator and / or a seat vibrator; wherein, the steering wheel vibrator includes a first LRA vibrator, and the seat vibrator includes a second LRA vibrator, the first LRA vibrator is embedded and installed in the left and right handle areas of the steering wheel, and the second LRA vibrator is embedded and installed in the position of the back of the driver's seat corresponding to the human lumbar area and / or the positions on both sides of the driver's seat corresponding to the human hip area.

6. The method according to claim 5, characterized in that The first LRA vibrator is set at a frequency of 20-50 Hz to deliver vibration stimulation at a subconscious level, thereby reminding the driver to concentrate; The setting frequency of the second LRA vibrator is 75-100 Hz to improve the driver's alertness during long-term driving or fatigue.

7. The method according to claim 6, characterized in that The set frequency of the first LRA vibrator is 25 Hz to match the sensitive frequency band of the Messner corpuscles in the hand; The setting frequency of the second LRA vibrator is 85 Hz to match the sensitive frequency band of the Pacinian corpuscles in the lower back.

8. The method according to claim 5, characterized in that The set intensities of the first LRA vibrator and the second LRA vibrator are both set to 0.05-0.1g, where g represents gravity acceleration; the set intensities of the first LRA vibrator and the second LRA vibrator are adjusted according to individual differences and feedback from drivers, and / or feedback and response data from drivers.

9. A distracted driving intervention system based on subthreshold vibration stimulation, characterized in that: The system comprises: A data acquisition module is configured to obtain eye movement data of the driver during driving; A data preprocessing module, configured to preprocess the eye movement data and extract eye movement features; A distraction prediction module is configured to classify the eye movement features according to a trained SVM model to determine whether the driver is distracted; A subthreshold vibration stimulation module is configured to start subthreshold vibration stimulation when it is determined that the driver is distracted, so that the vibrator works at a set frequency and a set intensity; Preprocessing the eye movement data and extracting eye movement features includes: Detecting and removing invalid data items in the eye movement data; wherein the invalid data items include null values ​​and data points whose coordinates exceed a set range; and based on a preset signal quality threshold, filtering the eye movement data, removing data points whose signal quality is lower than the preset signal quality threshold, and obtaining first eye movement data; Performing low-pass filtering on the first eye movement data to remove high-frequency noise, and adjusting the sampling frequency of the data to remove redundant high-frequency data, to obtain second eye movement data; When the eye movement coordinates in the second eye movement data exceed a set standard deviation threshold, determining that the eye movement coordinates in the second eye movement data are abnormal values, and correcting the abnormal values ​​based on linear interpolation to obtain third eye movement data; wherein the third eye movement data includes eye movement velocity and acceleration data; Based on the set speed threshold, if the eye movement speed is lower than the set speed threshold, it is identified as the fixation phase of the eye movement event; if the eye movement speed is not lower than the set speed threshold, it is identified as the scanning phase of the eye movement event; the duration of the fixation phase is determined according to the time period of the eye movement coordinate change, and the scanning speed is calculated from the rate of change of the eye movement trajectory; the duration of the fixation phase and the scanning speed are used as eye movement features.

Images