Driver driving style determination method and device, electronic equipment and storage medium

Abstract

The invention relates to the technical field of intelligent traffic, and provides a driver driving style determination method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining inertial measurement unit data of a vehicle; determining hazard acceleration behavior data based on the inertial measurement unit data; based on the driving instability index, classifying the average driving behavior of the driver to obtain a driving behavior classification result; determining a driving performance score and a risk index corresponding to the dangerous acceleration behavior based on the driving behavior classification result and the dangerous acceleration behavior data; and when the driving performance score is greater than a preset threshold value and the risk index is greater than a preset threshold value, extracting spatial-temporal characteristics of the inertial measurement unit data, and determining the driving style of the driver based on the spatial-temporal characteristics. The driving style of the driver is determined based on the spatio-temporal features, and the combination of the time features and the spatial features is combined, so that the driving behaviors are classified more finely, the most suitable auxiliary function is provided for the driver, and the driving experience and safety are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent transportation, and particularly relates to a method, device, electronic device and storage medium for determining a driver's driving style. Background Art

[0002] With the rapid development of the background technology of driving style identification, which stems from modern intelligent transportation systems and vehicle automation technologies, its core lies in accurately identifying the personalized driving habits of drivers to provide a customized basis for driving planning strategies. With the increasing requirements for driving experience and driving safety, driving style identification becomes particularly important. However, this task faces many challenges, including data diversity, complexity of feature selection, and model generalization ability, etc.

[0003] In recent years, thanks to the popularization of sensor technologies such as driving recorders and smartphones, the collection of large-scale natural driving data has become possible, providing rich materials for driving style analysis. In this context, researchers are committed to establishing an effective driving style identification model through key parameters such as acceleration, combined with machine learning and statistical analysis methods.

[0004] However, the recognition results of the driving style identification model for the driver's driving style are not fine enough, so that the most suitable auxiliary functions cannot be provided for the driver, thereby affecting the driving experience and safety. Summary of the Invention

[0005] The present invention provides a method, device, electronic device and storage medium for determining a driver's driving style, so as to solve the defect that the recognition results of the driving style identification model for the driver's driving style in the prior art are not fine enough, so that the most suitable auxiliary functions cannot be provided for the driver, thereby affecting the driving experience and safety.

[0006] The present invention provides a method for determining a driver's driving style, including the following steps: Obtain the inertial measurement unit data of the vehicle; Based on the inertial measurement unit data, determine the dangerous acceleration behavior data; Based on the driving instability index, classify the average driving behavior of the driver to obtain a driving behavior classification result; Based on the driving behavior classification result and the dangerous acceleration behavior data, determine the driving performance score and risk index corresponding to the dangerous acceleration behavior; When the driving performance score is greater than a first preset threshold and the risk index is greater than the first preset threshold, extract the spatio-temporal features of the inertial measurement unit data, and based on the spatio-temporal features, determine the driver's driving style.

[0007] A method for determining a driver's driving style provided by the present invention, the determining of dangerous acceleration behavior data based on the data of the inertial measurement unit includes: Based on the lateral acceleration data and longitudinal acceleration data in the data of the inertial measurement unit, determine a distribution map; Based on the data point density of each grid cell in the distribution map, estimate the data point distribution, and use the data points in the distribution map other than the data point distribution as the dangerous acceleration behavior data; The density is determined based on the total number of grid cells in the distribution map and the number of data points in each grid cell in the distribution map.

[0008] A method for determining a driver's driving style provided by the present invention, the estimating of the data point distribution based on the data point density of each grid cell in the distribution map includes: Sort the data point densities of each grid cell in the distribution map in descending order to obtain a sorting result; Use the grid cells corresponding to the top preset number of data point densities in the sorting result as the data point distribution.

[0009] A method for determining a driver's driving style provided by the present invention, the determining of the driving performance score and risk index corresponding to the dangerous acceleration behavior based on the driving behavior classification result and the dangerous acceleration behavior data includes: Based on the following formula, determine the driving performance score and risk index corresponding to the dangerous acceleration behavior: ; ; ; Wherein, represents the risk index, represents the percentage of the dangerous acceleration actions of driver m, | | is the modulus of the centroid of the risk component, wherein the centroid of the risk component is , and the modulus of the centroid of the risk component is ; represents the probability density function of the th Gaussian distribution, represents the sum value of the probability density functions, K is the total number of Gaussian distributions, and are the mean value and covariance matrix thereof respectively, is the probability that the weight of this Gaussian distribution corresponds to a certain classification in the driving behavior classification result for a certain dangerous driving behavior, and the probability here is related to the driver's dangerous acceleration behavior, rather than all acceleration behaviors; Indicates the driving performance score, Indicates the average risk index of the drivers in the safest group.

[0010] According to a method for determining a driver's driving style provided by the present invention, the extraction of the spatio-temporal features of the inertial measurement unit data includes: Extracting the spatial and temporal distributions of the dangerous acceleration behaviors and the individual behavior hot spot information in the inertial measurement unit data as the spatio-temporal features.

[0011] According to a method for determining a driver's driving style provided by the present invention, the individual behavior hot spot information is obtained from the target intersection positions where dangerous acceleration behaviors occur in the spatial and temporal distributions; The target intersection positions are the intersection positions where the frequency of dangerous driving behaviors at each intersection is greater than a second preset threshold; The determination step of the frequency is: Matching the intersection positions corresponding to the dangerous driving behaviors with the navigation map, and determining the frequency based on the number of intersection positions successfully matched in the navigation map and the total number of intersection positions.

[0012] According to a method for determining a driver's driving style provided by the present invention, the classification of the average driving behaviors of the driver based on the driving instability index to obtain the driving behavior classification result includes: Based on the driving instability index, determining the dissimilarity matrix between two clusters in the agglomerative method, and classifying the average driving behaviors of the driver based on the dissimilarity matrix to obtain the driving behavior classification result.

[0013] The present invention also provides a device for determining a driver's driving style, including the following units: An acquisition unit, configured to acquire the inertial measurement unit data of the vehicle; A first determination unit, configured to determine the dangerous acceleration behavior data based on the inertial measurement unit data; A classification unit, configured to classify the average driving behaviors of the driver based on the driving instability index to obtain the driving behavior classification result; A second determination unit, configured to determine the driving performance score and the risk index corresponding to the dangerous acceleration behavior based on the driving behavior classification result and the dangerous acceleration behavior data; A third determination unit, configured to extract the spatio-temporal features of the inertial measurement unit data and determine the driver's driving style based on the spatio-temporal features when the driving performance score is greater than a first preset threshold and the risk index is greater than the first preset threshold.

[0014] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the method for determining the driving style of a driver as described in any one of the above is implemented.

[0015] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method for determining the driving style of a driver as described in any one of the above is implemented.

[0016] The present invention also provides a computer program product, including a computer program. When the computer program is executed by a processor, the method for determining the driving style of a driver as described in any one of the above is implemented.

[0017] For the method, device, electronic device, and storage medium for determining the driving style of a driver provided by the present invention, on the one hand, when the driving performance score is greater than the first preset threshold and the risk index is greater than the first preset threshold, the spatio-temporal features of the inertial measurement unit data are extracted, which can effectively reduce the amount of data to be analyzed, make the subsequent analysis process more efficient, and save computing resources and time. On the other hand, based on the spatio-temporal features, the driving style of the driver is determined by combining the time features and the space features, so as to classify the driving behavior more precisely, and further provide the most suitable auxiliary functions for the driver, further improving the driving experience and safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0019] Figure 1 is one of the flowcharts of the method for determining the driving style of a driver provided by the present invention.

[0020] Figure 2 is the second flowchart of the method for determining the driving style of a driver provided by the present invention.

[0021] Figure 3 is the structural diagram of the device for determining the driving style of a driver provided by the present invention.

[0022] Figure 4 is the structural diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Apparently, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without creative efforts shall fall within the protection scope of the present invention.

[0024] The terms "first", "second", etc. in the present invention are used to distinguish similar objects and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are usually of the same type.

[0025] In the related art, in the field of driving style recognition, existing research and methods can be mainly divided into two categories: the classification method based on questionnaire surveys and the classification method based on in-vehicle experimental data. For the classification method based on questionnaire surveys, this method focuses on evaluating driving styles through questionnaires. Some widely recognized questionnaires include the Driver Behaviour Questionnaire (DBQ), the Driver Style Questionnaire (DSQ), and the Multi-dimensional Driving Style Inventory (MDSI). These questionnaires are designed to detect bad driving behaviors in self-reports and study the influence of different factors such as region, culture, age, and gender on the differences in driver behavior. However, the questionnaire form is highly subjective and relies on the driver's self-assessment, and it cannot accurately reflect the actual driving behavior. For the classification method based on in-vehicle experimental data, with the development of vehicle intelligence, researchers have begun to use the actual operating parameters of vehicles, such as speed, acceleration, etc., to construct a classification mechanism for driving styles. However, in most clustering analyses, the single K-means average clustering is used. Although K-means performs well in small-scale and simply distributed data, in the face of actual complex scenarios, its disadvantages limit its application scope. In terms of model application, some studies do not conduct detailed driving performance evaluations and risk component decompositions, and when evaluating the model, there is a lack of a verification link for the accuracy of driving performance scores using collision data, and the evaluation of model accuracy and reliability is not perfect.

[0026] Existing research on driving style recognition has some deficiencies, mainly reflected in the utilization of data, the interpretability of models, and the accuracy of identifying dangerous driving behaviors. Traditional methods often rely on single features and ignore the relationship between lateral and longitudinal accelerations, resulting in inaccurate driving style classification results. In addition, many studies use complex deep learning models, which, although having good performance, lack interpretability and are difficult to provide clear risk feedback and guidance to drivers. At the same time, the analysis of dangerous driving behaviors is usually based on fixed threshold judgments, without fully considering their sparsity in driving data, which may lead to the omission of high-risk behaviors. Moreover, the effective utilization of large-scale unlabeled data is a long-term challenge.

[0027] Based on the above problems, the present invention provides a method for determining a driver's driving style. Figure 1 It is one of the flow schematic diagrams of the method for determining a driver's driving style provided by the present invention. As Figure 1 shown, the method includes step 110, step 120, step 130, step 140, and step 150.

[0028] Step 110, obtain the inertial measurement unit data of the vehicle; Step 120, determine the dangerous acceleration behavior data based on the inertial measurement unit data; Step 130, classify the average driving behavior of the driver based on the driving instability index to obtain a driving behavior classification result; Step 140, determine the driving performance score and risk index corresponding to the dangerous acceleration behavior based on the driving behavior classification result and the dangerous acceleration behavior data; Step 150, when the driving performance score is greater than a first preset threshold and the risk index is greater than the first preset threshold, extract the spatio-temporal features of the inertial measurement unit data, and determine the driver's driving style based on the spatio-temporal features.

[0029] Specifically, first, obtain the data of the vehicle's Inertial Measurement Unit (IMU). The IMU data may include linear acceleration data, rotational angular velocity data, etc. The accelerometer in the IMU measures the linear acceleration of the vehicle along the three orthogonal axes of X, Y, and Z. These data can be used to calculate the speed and displacement of the vehicle. The gyroscope in the IMU measures the rotational angular velocity of the vehicle around the three orthogonal axes of X, Y, and Z. These data can be used to determine the attitude of the vehicle (pitch angle, roll angle, and yaw angle). For example, the gyroscope can detect the steering angle and rotation state of the vehicle. By integrating, the precise attitude of the vehicle can be obtained. By integrating the acceleration and angular velocity data, the pitch angle, roll angle, and yaw angle of the vehicle can be obtained. These attitude angles describe the direction and attitude of the vehicle in three-dimensional space.

[0030] After obtaining the IMU data, the dangerous acceleration behavior data can be determined based on the IMU data. Specifically, based on the lateral acceleration data and longitudinal acceleration data in the IMU data, a distribution map can be determined. Here, the distribution map is the G-G map.

[0031] It can be understood that the distribution map determined by combining the lateral acceleration and longitudinal acceleration is a framework for characterizing the driver's characteristics. This framework considers the relationship between the two acceleration axes to evaluate the driving style, thereby improving the accuracy and reliability of the subsequent determination of the driver's driving style.

[0032] Here, the G-G map is a chart used to evaluate the performance of road vehicles. It shows the changes in lateral acceleration and longitudinal acceleration of the vehicle under various driving conditions. Specifically, the G-G map uses the lateral acceleration data as the abscissa and the longitudinal acceleration data as the ordinate. Each data point is plotted in the coordinate system of the G-G map. The distribution of these data points constitutes the basic form of the G-G map. Through the G-G map, the dangerous acceleration behavior of the vehicle under different working conditions can be clearly shown, which helps to analyze the influence of vehicle pitch and roll on the suspension performance, and further evaluate the handling stability and dynamic performance of the vehicle.

[0033] Furthermore, based on the data point density of each grid cell in the distribution map, estimate the data point distribution, and regard the data points other than the data point distribution in the distribution map as the dangerous acceleration behavior data.

[0034] It can be understood that by using statistical methods to analyze the distribution of data points in the G-G diagrams of a large number of drivers, since dangerous driving behaviors are rare and multiple previous studies have regarded low-probability driving behaviors as dangerous driving behaviors, a 95% confidence region for the distribution of all drivers' acceleration data points can be defined and used as the safe driving region, which covers the acceleration operation ranges of most drivers under normal driving conditions. The specific steps are as follows: Divide the G-G diagram plane into grid cells of a fixed size, and use the density of data points in each grid cell to estimate the data point distribution. Among them, the density is determined based on the total number of grid cells in the distribution diagram and the number of data points in each grid cell of the distribution diagram. The formula for density is as follows: ; Among them, represents the density of data points in each grid cell in the grid cell, represents the number of data points in each grid cell of the distribution diagram, represents the th data point in the grid cell, C represents the total number of grid cells in the distribution diagram.

[0035] Then, sort the data point densities of each grid cell in the distribution diagram in descending order to obtain the sorting result. The formula is as follows: ; It can be understood that in order to represent the most common data points, that is, the data of the most common driving behaviors, it is necessary to estimate the confidence region of the distribution diagram.

[0036] q The confidence region corresponds to the cells where the first k densities are located, and k is calculated by the following formula: ; Among them, represents the data point density of the k th grid cell.

[0037] Correspondingly, the grid cells corresponding to the data point densities of the first preset number of data points in the sorting result are used as the data point distribution, so as to help us effectively distinguish dangerous driving behaviors and evaluate the aggressiveness or risk level of drivers.

[0038] Then, based on the driving instability index, classify the average driving behaviors of drivers to obtain the driving behavior classification result. Among them, the driving behavior classification result includes high risk, medium risk, and low risk.

[0039] Here, based on the hierarchical clustering algorithm and the driving instability index, the average driving behavior of the driver can be classified to obtain the driving behavior classification result.

[0040] Furthermore, based on the driving behavior classification result and the dangerous acceleration behavior data, the driving performance score and risk index corresponding to the dangerous acceleration behavior are determined.

[0041] For example, the driving behavior classification result and the dangerous acceleration behavior data are input into a Gaussian Mixture Model (GMM) to obtain the driving performance score and risk index corresponding to the dangerous acceleration behavior output by the Gaussian mixture model.

[0042] Here, the driving performance score is used to reflect the safety of the driver's overall driving behavior, while the risk index reveals potential dangerous driving behaviors. By analyzing these risk factors, the driving risk can be evaluated more accurately.

[0043] For example, based on the following formula, the driving performance score and risk index corresponding to the dangerous acceleration behavior are determined: ; ; ; where represents the risk index, represents the percentage of the dangerous acceleration actions of driver m, | | is the norm of the centroid of the risk component, where the centroid of the risk component is and the norm of the centroid of the risk component is ; represents the probability density function of the th Gaussian distribution, represents the sum value of the probability density functions, K is the total number of Gaussian distributions, and are its mean and covariance matrix respectively, is the probability that the weight of this Gaussian distribution corresponds to a certain dangerous driving behavior belonging to a certain classification in the driving behavior classification result. Here, the probability is related to the driver's dangerous acceleration behavior, rather than all acceleration behaviors; represents the driving performance score, represents the average risk index of the safest group of drivers.

[0044] It should be noted that the average risk index of the safest group of drivers is the average risk index of the safest group of drivers in the driving behavior classification results obtained by classifying the average driving behavior of drivers based on the driving instability index, that is, the average risk index of low-risk drivers in the driving behavior classification results.

[0045] It can be understood that in the embodiments of the present invention, an unsupervised method is adopted to model the driving behavior of drivers, making it applicable to the large-scale driving data that is becoming increasingly common in the context of intelligent transportation systems. In addition, a Gaussian mixture model is also adopted to evaluate the overall driving performance of drivers, and numerous dangerous driving behaviors are decomposed into several driving styles, thus facilitating driving behavior analysis.

[0046] After obtaining the driving performance score and the risk index, when the driving performance score is greater than the first preset threshold and the risk index is greater than the first preset threshold, the spatio-temporal features of the inertial measurement unit data can be extracted, and based on the spatio-temporal features, the driving style of the driver can be determined.

[0047] Here, the first preset threshold can be set to half of the sum of the driving performance score and the risk index, and can also be adjusted according to the actual situation. The embodiments of the present invention do not make specific limitations on this.

[0048] It should be noted that when both the driving performance score and the risk index are greater than the first preset threshold, it usually means that the driving behavior of the driver has relatively obvious abnormal or radical characteristics during this period. Extracting the spatio-temporal features of the inertial measurement unit data during these periods can more clearly capture the special patterns of the driver in terms of speed change, acceleration change, direction change, etc., providing a more powerful basis for the accurate identification of the driving style and avoiding inaccurate identification caused by the interference of normal driving behavior data.

[0049] Here, the driving style of the driver can include radical, steady, cautious, etc. The embodiments of the present invention do not make specific limitations on this.

[0050] It can be understood that the spatio-temporal features can include spatial features, time features, behavior frequency, and behavior intensity, etc. The spatial feature refers to the specific location where the dangerous acceleration behavior occurs, such as longitude, latitude, altitude, etc. The time feature refers to the specific time (date and time) when the dangerous acceleration behavior occurs. The behavior frequency refers to the number of times each driver has a dangerous acceleration behavior within a specific time period. The behavior intensity refers to the intensity of each dangerous acceleration behavior (the magnitude of the acceleration), etc.

[0051] It should be noted that in the embodiments of the present invention, the spatio-temporal features of the inertial measurement unit data are extracted, and based on the spatio-temporal features, the driving style of the driver is determined. The spatio-temporal features are used to reflect the spatio-temporal background of dangerous driving behaviors to improve the interpretability of the results obtained by machine learning algorithms.

[0052] Furthermore, spatio-temporal features can be used to create a driver style profile, and personalized feedback can be provided based on the driver style profile to improve driving safety.

[0053] It can be understood that spatial features depict the driver's behavioral habits and style characteristics in space from different angles, and temporal features can capture the dynamic changes in the driver's behavior, reflecting their driving decisions and behavioral patterns at different times, making the classification more accurate. The fusion of these two multi-dimensional spatial and temporal features can make up for the deficiencies of single-dimensional features, so that driving behaviors can be classified more precisely, avoiding classification errors caused by ignoring some important information.

[0054] The method provided by the embodiment of the present invention obtains the inertial measurement unit data of the vehicle, then determines the dangerous acceleration behavior data based on the inertial measurement unit data, classifies the average driving behavior of the driver based on the driving instability index to obtain a driving behavior classification result, and then determines the driving performance score and risk index corresponding to the dangerous acceleration behavior based on the driving behavior classification result and the dangerous acceleration behavior data. Finally, when the driving performance score is greater than the first preset threshold and the risk index is greater than the first preset threshold, the spatio-temporal features of the inertial measurement unit data are extracted, and the driver's driving style is determined based on the spatio-temporal features. On the one hand, when the driving performance score is greater than the first preset threshold and the risk index is greater than the first preset threshold, the method extracts the spatio-temporal features of the inertial measurement unit data, which can effectively reduce the amount of data to be analyzed, make the subsequent analysis process more efficient, and save computing resources and time. On the other hand, based on the spatio-temporal features, the driver's driving style is determined, and the combination of temporal features and spatial features enables more precise classification of driving behaviors, and further provides the most suitable auxiliary functions for drivers, further improving the driving experience and safety.

[0055] The method provided by the embodiment of the present invention classifies the driving style and quantifies the driving behavior performance by combining IMU data with the hierarchical clustering method, adapts to the individual differences of different drivers, and ensures the accuracy of the evaluation; introduces the Gaussian mixture model to finely model the dangerous driving mode, quantifies the risk index and reveals potential risk factors, providing clear guidance for optimizing driving behaviors. At the same time, it combines spatio-temporal feature analysis to locate high-risk areas and times, providing dynamic and real-time risk feedback. Therefore, the method provided by the embodiment of the present invention has the characteristics of low cost and high scalability, and is widely applicable to intelligent transportation and driving safety assessment scenarios.

[0056] It should be noted that the expansion or alternative solutions of the embodiments of the present invention can be realized through multi-faceted optimizations for wider applications. For example, multi-sensor fusion such as lidar, millimeter-wave radar, or GPS (Global Positioning System) can be introduced to enhance the perception ability in complex scenarios; deep learning or Bayesian networks can be used to replace the existing algorithms to improve the ability to capture non-linear features; real-time map updates and behavior prediction technologies can be combined to dynamically analyze driving behaviors and early warn of potential risks. At the same time, low-cost hardware replacement can be achieved by using smartphone sensors, or the computing tasks can be migrated to the cloud to reduce the requirements for terminal devices. In addition, this solution can be extended to fields such as autonomous driving, fleet management, and driver training, supporting personalized driving optimization, dynamic insurance pricing, and driving safety improvement, demonstrating strong flexibility and adaptability.

[0057] Based on the above embodiments, step 150 includes: Step 151, extracting the spatial and temporal distributions of the dangerous acceleration behavior and individual behavior hotspot information in the inertial measurement unit data as the spatio-temporal features.

[0058] Specifically, extract the spatial and temporal distributions of the dangerous acceleration behavior and individual behavior hotspot information in the inertial measurement unit data as spatio-temporal features.

[0059] It can be understood that the spatial and temporal distributions refer to spatial features and temporal features. The spatial features refer to the specific locations where the dangerous acceleration behavior occurs, such as longitude, latitude, altitude, etc., and the temporal features refer to the specific times (dates and times) when the dangerous acceleration behavior occurs, etc.

[0060] The individual behavior hotspot information is obtained from the target intersection locations where the dangerous acceleration behavior occurs in the spatial and temporal distributions.

[0061] The target intersection location is the intersection location where the frequency of dangerous driving behavior at each intersection is greater than a second preset threshold. The second preset threshold can be 50%, or can be adjusted based on the actual situation. The embodiments of the present invention do not make specific limitations on this.

[0062] Among them, the steps for determining the frequency are: Match the intersection locations corresponding to the dangerous driving behavior with the navigation map, and determine the frequency based on the number of intersection locations successfully matched in the navigation map and the total number of intersection locations, that is, determine the frequency of the dangerous driving behavior at each intersection as the number of locations matched with this intersection divided by the total number of intersection locations.

[0063] Based on the above embodiments, step 130 includes: Step 131: Based on the driving instability index, determine the dissimilarity matrix between two clusters in the agglomerative method, and classify the average driving behavior of the driver based on the dissimilarity matrix to obtain the driving behavior classification result.

[0064] Specifically, the driving instability index is calculated by the following formula: ; where, ( ) represents the th data point in the G-G plot, and respectively represent the lateral acceleration and longitudinal acceleration of the th data point, ( ) represents the centroid of all data points, represents the total number of data points.

[0065] It can be understood that the variable of the driving instability index is used to measure the stability of driving acceleration behavior. The larger the value of the driving instability index, the more unstable the driving is, and the smaller the value of the driving instability index, the more stable the driving is.

[0066] Cluster the average driving behavior through the agglomerative method. The agglomerative method initially treats each observation as a cluster and calculates the pairwise dissimilarity matrix between two clusters based on the linkage criterion. At each step, the two most similar clusters are identified and merged according to the dissimilarity matrix. By repeating this process, the clusters are merged into broader clusters until all observations are merged into one cluster. Among them, the dissimilarity matrix between two clusters is determined based on the driving instability index.

[0067] It should be noted that the number of clusters can be determined by the silhouette coefficient, so as to achieve the division of the average driving behavior. The silhouette coefficient is the average silhouette width of all data points in a cluster. The silhouette width of the data point can be obtained by the following formula: ; where, represents the silhouette width of the data point , is the average distance between the data point and all other data points in the same cluster, is the average distance between the data point and all data points in the nearest cluster.

[0068] It should be noted that the value range of the silhouette coefficient is between -1 and 1. A value close to 1 indicates that the data points are well clustered and assigned to appropriate clusters, while a value close to -1 indicates that the data points may be misclassified.

[0069] Finally, according to the data after clustering, the average driving behavior of the driver is divided into driving behavior classification results, and a risk level is marked for each classification, for example, high risk, medium risk, and low risk.

[0070] Based on the above embodiments, Figure 2 is the second schematic flowchart of the method for determining the driving style of a driver provided by the present invention. As Figure 2 shown, the method includes: First, obtain the data of the inertial measurement unit of the vehicle, that is, the data collected by the IMU. Then, based on hierarchical clustering and the driving instability index, classify the average driving behavior of the driver to obtain the driving behavior classification result. Then, combine the lateral acceleration and longitudinal acceleration data in the data collected by the IMU to determine the dangerous acceleration behavior data.

[0071] After obtaining the driving behavior classification result and the dangerous acceleration behavior data, the driving behavior classification result and the dangerous acceleration behavior data can be input into the Gaussian mixture model to obtain the driving performance score and risk index corresponding to the dangerous acceleration behavior output by the Gaussian mixture model.

[0072] It should be noted that the driving behavior classification result is the benchmark of the driving performance score, and the risk index is the supplementary result of the dangerous acceleration behavior data.

[0073] When the driving performance score is greater than the first preset threshold and the risk index is greater than the first preset threshold, extract the spatio-temporal features of the inertial measurement unit data, and based on the spatio-temporal features, determine the driving style of the driver.

[0074] It should be noted that the spatio-temporal features can provide implicit factors of the driver's driving behavior, so as to analyze the driver's risk situation.

[0075] Next, the driver driving style determination device provided by the present invention will be described. The driver driving style determination device described below can be correspondingly referred to the driver driving style determination method described above.

[0076] Based on any of the above embodiments, the present invention provides a driver driving style determination device, Figure 3 is the structural schematic diagram of the driver driving style determination device provided by the present invention. As Figure 3 shown, the device includes: An acquisition unit 310, configured to acquire the data of the inertial measurement unit of the vehicle; The first determination unit 320 is configured to determine dangerous acceleration behavior data based on the inertial measurement unit data; The classification unit 330 is configured to classify the average driving behavior of the driver based on the driving instability index to obtain a driving behavior classification result; The second determination unit 340 is configured to determine a driving performance score and a risk index corresponding to the dangerous acceleration behavior based on the driving behavior classification result and the dangerous acceleration behavior data; The third determination unit 350 is configured to extract spatio-temporal features of the inertial measurement unit data and determine the driving style of the driver based on the spatio-temporal features when the driving performance score is greater than a first preset threshold and the risk index is greater than the first preset threshold.

[0077] On the one hand, the device provided by the embodiment of the present invention, when the driving performance score is greater than the first preset threshold and the risk index is greater than the first preset threshold, extracts the spatio-temporal features of the inertial measurement unit data, which can effectively reduce the amount of data to be analyzed, make the subsequent analysis process more efficient, and save computing resources and time. On the other hand, based on the spatio-temporal features, the driving style of the driver is determined, and the combination of the time feature and the space feature is used to classify the driving behavior more precisely, and then provide the most suitable auxiliary function for the driver, further improving the driving experience and safety.

[0078] Based on any of the above embodiments, the first determination unit 320 specifically includes: The determination distribution map unit is configured to determine a distribution map based on the lateral acceleration data and the longitudinal acceleration data in the inertial measurement unit data; The estimation unit is configured to estimate the data point distribution based on the data point density of each grid unit in the distribution map, and use the data points in the distribution map except the data point distribution as the dangerous acceleration behavior data; The density is determined based on the total number of grid units in the distribution map and the number of data points in each grid unit in the distribution map.

[0079] Based on any of the above embodiments, the estimation unit is specifically configured to: Sort the data point densities of each grid unit in the distribution map in descending order to obtain a sorting result; Use the grid units corresponding to the top preset number of data point densities in the sorting result as the data point distribution.

[0080] Based on any of the above embodiments, the second determination unit 340 is specifically configured to: Determine the driving performance score and the risk index corresponding to the dangerous acceleration behavior based on the following formula: ; ; ; Among them, represents the risk index, represents the percentage of dangerous acceleration actions of driver m, | | is the modulus length of the centroid of the risk component, where the centroid of the risk component is , and the modulus length of the centroid of the risk component is ; represents the probability density function of the th Gaussian distribution, represents the sum value of the probability density function, K is the total number of Gaussian distributions, and are its mean and covariance matrix respectively, is the probability that the weight of this Gaussian distribution corresponds to a certain classification in the driving behavior classification result for a certain dangerous driving behavior. The probability here is related to the driver's dangerous acceleration behavior, rather than all acceleration behaviors; represents the driving performance score, represents the average risk index of the safest group of drivers.

[0081] Based on any of the above embodiments, the third determination unit 350 specifically includes: A spatio-temporal feature extraction unit for extracting the spatial and temporal distributions of the dangerous acceleration behavior and the individual behavior hot spot information in the inertial measurement unit data as the spatio-temporal features.

[0082] Based on any of the above embodiments, the individual behavior hot spot information is obtained from the target intersection positions where dangerous acceleration behaviors occur in the spatial and temporal distributions; The target intersection position is the intersection position where the frequency of dangerous driving behaviors at each intersection is greater than a second preset threshold; It further includes a frequency determination unit, and the frequency determination unit is specifically used for: Matching the intersection positions corresponding to the dangerous driving behaviors with the navigation map, and determining the frequency based on the number of intersection positions successfully matched in the navigation map and the total number of intersection positions.

[0083] Based on any of the above embodiments, the classification unit 330 is specifically used for: Based on the driving instability index, determining the dissimilarity matrix between two clusters in the agglomerative method, and classifying the average driving behavior of the driver based on the dissimilarity matrix to obtain the driving behavior classification result.

[0084] Figure 4 is a schematic structural diagram of the electronic device provided by the present invention. As Figure 4 shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430, and a communication bus 440. Among them, the processor 410, the communication interface 420, and the memory 430 complete communication with each other through the communication bus 440. The processor 410 may call the logical instructions in the memory 430 to execute the method for determining the driving style of a driver, and the method includes: obtaining inertial measurement unit data of a vehicle; determining dangerous acceleration behavior data based on the inertial measurement unit data; classifying the average driving behavior of the driver based on a driving instability index to obtain a driving behavior classification result; determining a driving performance score and a risk index corresponding to the dangerous acceleration behavior based on the driving behavior classification result and the dangerous acceleration behavior data; and when the driving performance score is greater than a first preset threshold and the risk index is greater than the first preset threshold, extracting spatio-temporal features of the inertial measurement unit data and determining the driving style of the driver based on the spatio-temporal features.

[0085] In addition, when the logical instructions in the above-mentioned memory 430 can be implemented in the form of a software functional unit and sold or used as an independent product, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk, or an optical disc that can store program codes.

[0086] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the driver driving style determination method provided by the above-mentioned various methods. The method includes: obtaining inertial measurement unit data of a vehicle; determining dangerous acceleration behavior data based on the inertial measurement unit data; classifying the average driving behavior of the driver based on the driving instability index to obtain a driving behavior classification result; determining a driving performance score and a risk index corresponding to the dangerous acceleration behavior based on the driving behavior classification result and the dangerous acceleration behavior data; when the driving performance score is greater than a first preset threshold and the risk index is greater than the first preset threshold, extracting spatio-temporal features of the inertial measurement unit data, and determining the driver's driving style based on the spatio-temporal features.

[0087] In another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it realizes the driver driving style determination method provided by the above-mentioned various methods. The method includes: obtaining inertial measurement unit data of a vehicle; determining dangerous acceleration behavior data based on the inertial measurement unit data; classifying the average driving behavior of the driver based on the driving instability index to obtain a driving behavior classification result; determining a driving performance score and a risk index corresponding to the dangerous acceleration behavior based on the driving behavior classification result and the dangerous acceleration behavior data; when the driving performance score is greater than a first preset threshold and the risk index is greater than the first preset threshold, extracting spatio-temporal features of the inertial measurement unit data, and determining the driver's driving style based on the spatio-temporal features.

[0088] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without creative labor.

[0089] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0090] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements 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.

Claims

1. A method for determining a driver's driving style, characterized in that: include: Obtain the vehicle's inertial measurement unit data; determining dangerous acceleration behavior data based on the inertial measurement unit data; Based on the driving instability index, the average driving behavior of the driver is classified to obtain the driving behavior classification result; Determining a driving performance score and a risk index corresponding to the dangerous acceleration behavior based on the driving behavior classification result and the dangerous acceleration behavior data; When the driving performance score is greater than a first preset threshold and the risk index is greater than the first preset threshold, spatiotemporal features of the inertial measurement unit data are extracted, and the driver's driving style is determined based on the spatiotemporal features.

2. The method for determining a driver's driving style according to claim 1, characterized in that: The step of determining the dangerous acceleration behavior data based on the inertial measurement unit data includes: Determining a distribution map based on the lateral acceleration data and the longitudinal acceleration data in the inertial measurement unit data; Based on the data point density of each grid cell in the distribution map, estimating the data point distribution, and using the data points in the distribution map other than the data point distribution as the dangerous acceleration behavior data; The density is determined based on the total number of grid cells in the distribution map and the number of data points of each grid cell in the distribution map.

3. The method for determining a driver's driving style according to claim 2, characterized in that: The estimating the data point distribution based on the data point density of each grid cell in the distribution map includes: Sorting the density of data points of each grid cell in the distribution map in descending order to obtain a sorting result; The grid units corresponding to the first preset number of data point densities in the sorting result are used as the data point distribution.

4. The method for determining a driver's driving style according to any one of claims 1 to 3, characterized in that: The determining, based on the driving behavior classification result and the dangerous acceleration behavior data, a driving performance score and a risk index corresponding to the dangerous acceleration behavior includes: The driving performance score and risk index corresponding to the dangerous acceleration behavior are determined based on the following formula: ； ； ； in, represents the risk index, represents the percentage of dangerous acceleration actions of driver m, | | is the modulus of the centroid of the risk component, where the centroid of the risk component is , the modulus of the centroid of the risk component is ; Indicates The probability density function of a Gaussian distribution is represents the sum of the probability density function, K is the total number of Gaussian distributions, and are their mean and covariance matrices, respectively, The weight of the Gaussian distribution corresponds to the probability that a certain dangerous driving behavior belongs to a certain category in the driving behavior classification result. The probability here is related to the driver's dangerous acceleration behavior, not all acceleration behaviors; represents the driving performance score, Represents the average risk index of the safest group of drivers.

5. The method for determining a driver's driving style according to any one of claims 1 to 3, characterized in that: The extracting the spatiotemporal features of the inertial measurement unit data comprises: The spatial and temporal distribution of the dangerous acceleration behavior and the individual behavior hot spot information in the inertial measurement unit data are extracted as the spatiotemporal features.

6. The method for determining a driver's driving style according to claim 5, characterized in that: The individual behavior hotspot information is obtained by displaying the target intersection location where the dangerous acceleration behavior occurs in the spatial and temporal distribution; The target intersection position is an intersection position where the frequency of dangerous driving behaviors at each intersection is greater than a second preset threshold; The steps for determining the frequency are: The intersection position corresponding to the dangerous driving behavior is matched with the navigation map, and the frequency is determined based on the number of successfully matched intersection positions and the total number of intersection positions in the navigation map.

7. The method for determining a driver's driving style according to any one of claims 1 to 3, characterized in that: The average driving behavior of the driver is classified based on the driving instability index to obtain a driving behavior classification result, including: Based on the driving instability index, a dissimilarity matrix between two clusters in a cohesive method is determined, and based on the dissimilarity matrix, the average driving behavior of the driver is classified to obtain the driving behavior classification result.

8. A device for determining a driver's driving style, characterized in that: include: An acquisition unit, used for acquiring inertial measurement unit data of the vehicle; A first determining unit, configured to determine dangerous acceleration behavior data based on the inertial measurement unit data; A classification unit, used for classifying the average driving behavior of the driver based on the driving instability index to obtain a driving behavior classification result; A second determination unit is used to determine a driving performance score and a risk index corresponding to the dangerous acceleration behavior based on the driving behavior classification result and the dangerous acceleration behavior data; The third determination unit is used to extract the spatiotemporal features of the inertial measurement unit data when the driving performance score is greater than a first preset threshold and the risk index is greater than the first preset threshold, and determine the driver's driving style based on the spatiotemporal features.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the method for determining the driver's driving style according to any one of claims 1 to 7 is implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for determining a driver's driving style as claimed in any one of claims 1 to 7 is implemented.

Images