Intelligent driving vehicle personification evaluation method and system

Abstract

The invention discloses an intelligent driving vehicle personification evaluation method and system, and relates to the technical field of intelligent driving, and the method comprises the steps: obtaining human driving characteristic parameters; clustering the feature parameters to obtain multiple groups of driving data; analyzing the plurality of groups of data to obtain a plurality of driving styles; classifying driving styles of the intelligent driving vehicle data; and calculating to obtain a personification score of the intelligent driving vehicle. According to the invention, through the principal component analysis method and the K-means + + algorithm, the driving data of the human driver in the target scene is clustered and the driving style classification is carried out, then through the K-means + + algorithm, the intelligent driving data is clustered, and the clustering result is matched to the driving style. The final personification score is obtained according to the classification of the driving styles and the illegal conditions in the driving process, and the personification evaluation is high in accuracy and standardization and is not easily interfered by subjective factors.

Description

Technical Field

[0001] The present application relates to the field of intelligent driving technology, and more specifically, to a method and system for anthropomorphic evaluation of an intelligent driving vehicle. Background Art

[0002] The Chinese national standard "Automobile Driving Automation Classification" (GB / T 40429-2021), which was implemented in March 2022, divides autonomous driving technology into 6 levels: Level 0 driving automation (emergency assistance), Level 1 driving automation (partial driving assistance), Level 2 driving automation (combined driving assistance), Level 3 driving automation (conditional autonomous driving), Level 4 driving automation (highly autonomous driving), and Level 5 driving automation (fully autonomous driving). With the rapid development of autonomous driving technology, the penetration rate of Level 2 autonomous driving has exceeded 50% so far, and some technologies are close to the intelligent level of Level 3 driving automation. my country's current car ownership has reached 150 million vehicles, and there will be a long-term situation where intelligent driving vehicles and driving vehicles are driving on the road at the same time. The driving decision of the driving vehicle is constantly changing according to the traffic conditions. Drivers can reach a certain tacit understanding with the surrounding vehicles based on the actual situation, but this is quite difficult for intelligent driving vehicles. If the driving behavior of the intelligent driving vehicle is greatly different from that of the driving vehicle, it will not only cause psychological panic to the driver and affect traffic efficiency, but also may bring safety risks. Therefore, making smart driving vehicles closer to the driver's driving pattern is crucial to improving traffic safety and traffic efficiency.

[0003] At present, there are few studies on the evaluation methods of the anthropomorphism of intelligent driving vehicles. There are mainly two methods: (1) scoring the safety, comfort, trustworthiness and other aspects of the vehicle through subjective scoring. This evaluation method is easily interfered by subjective factors, and there is no quantitative and clear standard to distinguish the degree of anthropomorphism; (2) by comparing the driving time with the selected excellent drivers in the same period and the same section of road. This method can reflect the degree of anthropomorphism to a certain extent, but the evaluation dimension is relatively single, so the evaluation of the degree of anthropomorphism is not accurate enough.

[0004] Therefore, in order to solve the above problems, it is necessary for us to design a reasonable anthropomorphic evaluation method and system for intelligent driving vehicles. Summary of the invention

[0005] The present invention provides an anthropomorphic evaluation method and system for intelligent driving vehicles. The driving data of the driver in the target scene is clustered and the human driving style is classified through principal component analysis and K-means++ algorithm. The intelligent human driving data is clustered through K-means++ algorithm, and the clustering result is matched to the human driving style. The final anthropomorphic score is obtained according to the classification of human driving style and the illegal and irregular situations during driving. The anthropomorphic evaluation is accurate and standardized, and is not easily interfered by subjective factors.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions: In a first aspect, an embodiment of the present application provides a method for anthropomorphic evaluation of an intelligent driving vehicle, the method comprising: S100: Acquire characteristic parameters of vehicle driving in a target scene, and select key characteristic parameters; S200: clustering the selected key parameters to obtain Q groups of human driving data; wherein Q is not less than 3; S300: Calculate the median of the speed and the median of the acceleration of the vehicle passing through the target scene in the Q groups of data, and correspond the Q groups of human driving data to Q types of human driving styles; S400: clustering the data of the intelligent driving vehicle in the target scenario, calculating the multiple groups of clustered data and comparing them with the Q types of human driving styles obtained in step S300, and then classifying and matching them; The classification matching method is: calculating the median of the speed and the median of the acceleration of the clustered multiple groups of data passing through the target scene, and comparing them with the median of the speed and the median of the acceleration of the vehicle passing through the target scene of any one of the Q human driving styles in step S300, and if the difference is less than a preset value, it is classified into the human driving style; S500: Calculate the anthropomorphic score of the intelligent driving vehicle according to the correction coefficient of the human driving style of the intelligent driving and the illegal situations of the intelligent driving during driving.

[0007] As a preferred embodiment of the present invention, when step S100 is performed, specifically: S101: Obtain m representative characteristic parameters of vehicle driving in a target scene; S102: reducing the dimensions of the m characteristic parameters by using the PCA principal component analysis method, calculating the contribution rate of each characteristic parameter, and sorting them from large to small; S103: When the sum of the contribution rates of the first n characteristic parameters is not less than a preset threshold, the first n characteristic parameters are selected as key parameters.

[0008] As a preferred embodiment of the present invention, when executing step S101, the characteristic parameters include: average vehicle speed, maximum vehicle speed, standard deviation of vehicle speed, average vehicle lateral acceleration, maximum vehicle lateral acceleration, standard deviation of vehicle lateral acceleration, average vehicle longitudinal acceleration, maximum vehicle longitudinal acceleration, standard deviation of vehicle longitudinal acceleration, number of surrounding vehicles when passing through an intersection, and time passing through a target scene.

[0009] As a preferred embodiment of the present invention, when executing step S200, specifically: the number of centroids is set to Q, the selected key parameters are clustered by the K-means++ algorithm, and finally Q groups of human driving data are obtained.

[0010] As a preferred embodiment of the present invention, when executing step S300, the median of the speed and the median of the acceleration of the vehicle passing through the target scene in the Q group data are calculated and compared, and the data are arranged in order from small to large, and the Q group data is classified into human driving styles; Among them: the lower median speed and median acceleration are conservative human driving styles, the medium median speed and median acceleration are stable human driving styles, and the larger median speed and median acceleration are aggressive human driving styles.

[0011] As a preferred embodiment of the present invention, when executing step S400, specifically: S401: Acquire driving data of an intelligent driving vehicle in a target scenario as data to be clustered; the driving data includes m characteristic parameters and n key parameters of the intelligent driving vehicle in the target scenario; S402: clustering the data to be clustered using the K-means++ algorithm to generate k groups of data; The number of centroids ranges from 2 to 9. The Calinski-Harabasz index under different numbers of centroids is calculated, and the number of centroids k when the Calinski-Harabasz index is the largest is selected. S403: Compare the k groups of data generated by clustering with the Q types of human driving styles obtained in step S300 and perform classification matching, and classify the k groups of data into the human driving styles respectively.

[0012] As a preferred embodiment of the present invention, when executing step S403, If the difference between any of the k groups of data and two of the Q types of human driving styles is smaller than a preset value, the group of data is classified into the human driving style with the smaller difference; If the difference between any of the k groups of data and the Q types of human driving style data is not less than a preset value, the group of data is classified as other human driving styles.

[0013] As a preferred embodiment of the present invention, before executing step S500, a correction coefficient of the human driving style is assigned to each type of human driving style in the Q types of human driving style data and other human driving styles.

[0014] As a preferred embodiment of the present invention, when executing step S500, specifically: S501: Obtaining a penalty point value L for illegal driving during driving; S502: Calculate the anthropomorphic score G of the intelligent driving vehicle according to the intelligent driving classification style; the anthropomorphic score calculation method is: G = (DL) * e * 100%; where D is the full score of the anthropomorphic evaluation; e is the correction coefficient of the human driving style corresponding to the intelligent driving during driving In a second aspect, an embodiment of the present application provides an anthropomorphic evaluation system for an intelligent driving vehicle, the system comprising: Control data feature acquisition module; Control data clustering module; Human driving style classification module; Intelligent driving style matching module; Anthropomorphic scoring module; The control data feature acquisition module obtains the characteristic parameters of human-driven vehicles in the target scene and selects key characteristic parameters; the control data clustering module clusters the selected key parameters to obtain Q groups of human driving data, where Q is not less than 3; the human driving style classification module calculates the median of the speed and the median of the acceleration of the vehicle passing through the target scene in the Q groups of human driving data, and corresponds the Q groups of human driving data to Q types of human driving styles; the intelligent driving style matching module clusters the data of the intelligent driving vehicle in the target scene, calculates the clustered multiple groups of data and compares them with the Q types of human driving styles for classification and matching; the classification and matching method is: calculate the median of the speed and the median of the acceleration of the clustered multiple groups of data passing through the target scene, and compares them with the median of the speed and the median of the acceleration of the vehicle passing through the target scene of any human driving style in the Q types of human driving styles. If the difference is less than the preset value, it is classified into the human driving style; the anthropomorphic scoring module calculates the anthropomorphic score of the intelligent driving vehicle according to the correction coefficient of the human driving style corresponding to the intelligent driving and the illegal deduction value of the intelligent driving during driving.

[0015] In a third aspect, an embodiment of the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method provided in any possible implementation manner of the first aspect is implemented.

[0016] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the method provided in any possible implementation manner of the first aspect is implemented.

[0017] The beneficial effects of the present invention are as follows: the present invention provides an anthropomorphic evaluation method and system for intelligent driving vehicles, which clusters the driving data of the driver in the target scene and classifies the human driving style through principal component analysis and K-means++ algorithm, and then clusters the intelligent human driving data through the K-means++ algorithm, matches the clustering results to the human driving style, and obtains the final anthropomorphic score according to the classification of human driving style and the illegal and irregular situations during driving. The anthropomorphic evaluation is accurate and standardized, and is not easily interfered by subjective factors. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0019] Figure 1 A flowchart of a method for anthropomorphic evaluation of an intelligent driving vehicle provided in an embodiment of the present application. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0021] In the following introduction, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance. The following introduction provides multiple embodiments of the present application, and different embodiments can be replaced or combined, so the present application can also be considered to include all possible combinations of the same and / or different embodiments recorded. Therefore, if one embodiment includes features A, B, and C, and another embodiment includes features B and D, then the present application should also be considered to include embodiments containing one or more of all other possible combinations of A, B, C, and D, although the embodiment may not be clearly recorded in the following text.

[0022] The following description provides examples and does not limit the scope, applicability or examples set forth in the claims. Changes may be made to the functions and arrangements of the elements described without departing from the scope of the present application. Various processes or components may be appropriately omitted, substituted or added to each example. For example, the described method may be performed in an order different from the order described, and various steps may be added, omitted or combined. In addition, features described in some examples may be combined in other examples.

[0023] Example 1: See Figure 1 , Figure 1 : is a flow chart of a method for anthropomorphic evaluation of an intelligent driving vehicle provided in an embodiment of the present application. In an embodiment of the present application, the method includes: S100: Acquire characteristic parameters of vehicle driving in a target scene, and select key characteristic parameters; The target scenarios include going straight on a straight road, changing lanes on a straight road, turning on a curve, avoiding pedestrians at an intersection, going straight at an intersection, turning left at an intersection, turning right at an intersection, etc. Each different target scenario can be scored using the anthropomorphic evaluation method of this application.

[0024] Since intersections are typical and complex traffic scenes on urban roads, vehicles need to deal with complex behavioral interactions of surrounding vehicles, non-motor vehicles, and pedestrians when passing through intersections. Intersections also have high requirements for the degree of anthropomorphism of intelligent driving vehicles when passing through them. Turning vehicles at intersections are more complicated than going straight, and the length of the formal route for vehicles turning left at intersections and the number of vehicles considered are greater than the length of the formal route for vehicles turning right at intersections and the number of vehicles considered. Therefore, this application takes the anthropomorphic evaluation method for the left turn scenario of the main vehicle at an intersection as an example, and all the data below are based on the left turn data of vehicles at an intersection.

[0025] The anthropomorphic evaluation methods for other scenes are the same, but the representative feature parameters of different scenes will be different and need to be selected according to actual conditions.

[0026] In the present application, the methods for acquiring the characteristic parameters of vehicle driving in the target scene include: cameras and drones. Here, the camera often has a certain inclination angle and the camera height is not enough. The urban intersection data taken by drones contains rich vehicle behavior data, such as speed, acceleration, interaction with surrounding vehicles, vehicle driving trajectory, target type, etc., which are very useful for analyzing the behavior patterns and characteristics of people driving vehicles when passing through intersections; therefore, in order to ensure data accuracy, this application takes drone aerial photography as an example to obtain the characteristic parameters of vehicle driving in the target scene.

[0027] In this embodiment, when step S100 is executed, specifically: S101: Acquire m representative characteristic parameters of a driving vehicle in a target scene; The characteristic parameters of the scene where the main vehicle turns left at the intersection include: the average vehicle speed v mean 、 Maximum vehicle speed v max , vehicle speed standard deviation v std 、The average lateral acceleration of the vehicle ax mean 、The maximum lateral acceleration of the vehicle ax max 、Vehicle lateral acceleration standard deviation ax std , average longitudinal acceleration of the vehicle ay mean 、The maximum longitudinal acceleration of the vehicle ay max 、Vehicle longitudinal acceleration standard deviation ay std , the number of surrounding vehicles p when passing the intersection and the time t to pass the intersection, so m is 11, then the feature parameter dimension is 11; Taking the left turn at an intersection as an example, the vehicle passing through the target scene, i.e., the intersection, is divided into three stages. In the first stage, the vehicle enters the intersection until the front of the vehicle touches the start and end lines of the road before turning; in the second stage, the vehicle turns left, i.e., the front of the vehicle touches the start and end lines of the road before turning, until the rear of the vehicle leaves the start and end lines of the road after turning; in the third stage, the vehicle leaves the intersection, i.e., after the rear of the vehicle completely leaves the start and end lines of the road after turning. The driving characteristic parameters of the target scene's driving vehicle are mainly for the driving parameters of the second stage.

[0028] The number of surrounding vehicles p involved in this application when passing through an intersection refers to vehicles in the same lane or adjacent lane as the main vehicle, and whose longitudinal distance does not exceed the length of the vehicle; the time to pass the intersection t refers to the driving time of the main vehicle in the target scene of the intersection (the second stage when turning left at the intersection); S102: reducing the dimensions of the m characteristic parameters by using the PCA principal component analysis method, calculating the contribution rate of each characteristic parameter, and sorting them from large to small; Here, the ‌‌Z-Score normalization method is used to process the parameters before dimensionality reduction, converting the 11 feature parameters to the same magnitude to avoid the impact of different parameter magnitudes; The PCA principal component analysis method is used to reduce the dimension of m (11) standardized parameters, calculate the contribution rate of each characteristic parameter, and sort them from large to small; S103: When the sum of the contribution rates of the first n characteristic parameters is not less than a preset threshold, the first n characteristic parameters are selected as key parameters.

[0029] The first n feature parameters are selected as key parameters to characterize the data characteristics of the original m-dimensional (11-dimensional) feature parameters, and it is determined whether the contribution rate of the feature parameter with the largest contribution rate is not less than the preset threshold. If so, only one feature parameter is output as the key parameter; otherwise, the contribution rates of multiple feature parameters are sorted from large to small, and the contribution rates are added in sequence until the sum of the contribution rates of multiple feature parameters is not less than the preset threshold, and these feature parameters are selected as key parameters; the number of key parameters is n.

[0030] Here, the sum of the contribution rates of the m feature parameters is 100%.

[0031] For example, the contribution rates of the 11 characteristic parameters are 21%, 15%, 12%, 11%, 10%, 9%, 8%, 5%, 4%, 3% and 2% from large to small, and the preset threshold requirement is 90%. Then, the contribution rates of the characteristic parameters with the largest contribution rates are added together, and the contribution rates of the top 7 characteristic parameters are only 86%, which is not less than the preset threshold. Therefore, the contribution rates of the top 8 characteristic parameters are continued to be accumulated, so that the contribution rates of the top 8 characteristic parameters are added together to obtain 91%, which is greater than the preset threshold of 90%. At this time, the requirement is met, so the value of n is 8.

[0032] S200: clustering the selected key parameters to obtain Q groups of human driving data; wherein Q is not less than 3; Here, the number of centroids is set to Q, and the selected key parameters are clustered through the K-means++ algorithm, and finally Q groups of human driving data are obtained. That is, the number of centroids during clustering is Q, and Q groups of human driving data can be clustered.

[0033] According to the survey, the driving styles of human drivers can be divided into three categories: conservative, stable and aggressive. Therefore, the value of the number of centroids is set to 3, and finally 3 groups of human driving data are obtained.

[0034] It should be noted that with the development of technology and the improvement of intelligent driving accuracy, the driving styles of human drivers can be further subdivided, which can be recorded as: severely conservative, mildly conservative, stable, mildly aggressive and severely aggressive. The value of the number of centroids is set to 5, and finally 5 groups of human driving data are obtained; of course, it can be further subdivided, for example: the first conservative type, the second conservative type... the αth conservative type, the first stable type, the second stable type... the βth stable type, the first aggressive type, the second aggressive type... the γth aggressive type, then the number of centroids and the Q value are equal to (α+β+γ).

[0035] We take three types of human driving styles as examples for calculation. Specifically, we use the K-means++ algorithm to cluster the key parameters after dimensionality reduction to identify the human driving style of the driver, and further analyze the driving behavior characteristics of different human driving styles. Calculate the median intersection speed v of the three groups of data med and the median acceleration a med , and the three groups of data are classified as conservative from low to high (v med—1 , a med_1 )、Stable type(v med—2 , a med_2 ) and aggressive type (v med—3 , a med_3 ).

[0036] S300: Calculate the median of the speed and the median of the acceleration of the vehicle passing through the target scene in the Q groups of human driving data, and correspond the Q groups of human driving data to Q types of human driving styles; The median is calculated for the speed and acceleration of the vehicle passing through the target scene in all human driving data in each of Q groups of human driving data, and the median of Q speeds and the median of acceleration are obtained. The median of the Q speeds and the median of acceleration are compared and arranged in order from small to large. At the same time, the Q driving styles are also arranged from conservative to aggressive, and the median of the Q speeds and the median of acceleration are mapped one by one to the Q driving styles.

[0037] When executing step S300, the median speed and the median acceleration of the vehicles passing through the target scene in the Q group data are calculated and compared, and arranged in order from small to large, and the Q group data is classified into human driving styles; wherein: the lower the median speed and the median acceleration are, the conservative human driving style is; the medium the median speed and the median acceleration are, the stable human driving style is; the larger the median speed and the median acceleration are, the aggressive human driving style is.

[0038] Let's take three human driving styles (conservative, smooth and aggressive) as an example for calculation. That is, there are three groups of human driving data. The median speed and median acceleration of the three groups of data passing through the intersection are compared. The lowest is conservative, the middle is smooth, and the highest is aggressive. For example, in the first group of human driving data, the median speed and median acceleration of the vehicle passing through the target scene are 4m / s and 3m / s respectively. 2 , The median speed and acceleration of the vehicle passing through the target scene in the second set of human driving data are 7m / s and 6m / s respectively. 2 , The median speed and acceleration of the vehicle passing through the target scene in the third group of human driving data are 6m / s and 5m / s respectively 2, arranged from small to large, respectively, as the first group, the third group and the second group. Then after corresponding to the three human driving styles, we can get: the first group of human driving data corresponds to the conservative type, the third group of human driving data corresponds to the stable type, and the second group of human driving data corresponds to the aggressive type.

[0039] It can be understood that steps S100 to S300 are to obtain and calculate the driving data of human-driven vehicles as reference data for the driving data of intelligent driving vehicles, so as to facilitate the anthropomorphic scoring of intelligent driving data.

[0040] Step S400 starts with analyzing the driving data of the intelligent driving vehicle (which can be referred to as intelligent driving data for short).

[0041] S400: clustering the data of the intelligent driving vehicle in the target scenario, calculating the multiple groups of clustered data and comparing them with the Q types of human driving styles obtained in step S300, and then classifying and matching them; The classification matching method is: calculating the median of the speed and the median of the acceleration of the clustered multiple groups of data passing through the target scene, and comparing them with the median of the speed and the median of the acceleration of the vehicle passing through the target scene of any one of the Q human driving styles in step S300, and if the difference is less than a preset value, it is classified into the human driving style; In the present invention, when step S400 is executed, specifically: S401: Acquire driving data of an intelligent driving vehicle in a target scenario as data to be clustered; the driving data includes m characteristic parameters and n key parameters of the intelligent driving vehicle in the target scenario; S402: clustering the clustered data using the K-means++ algorithm to generate k groups of intelligent driving data; The number of centroids ranges from 2 to 9. The Calinski-Harabasz index under different numbers of centroids is calculated, and the number of centroids k when the Calinski-Harabasz index is the largest is selected. Here, the Calinski-Harabasz index is calculated when the number of centroids is 2, 3, 4, 5, 6, 7, 8, and 9, respectively. The Calinski-Harabasz index (abbreviated as CH) determines the clustering effect of different number of centroids k: B is the inter-cluster dispersion, which indicates the distance between different cluster centers; W is the intra-cluster dispersion, which indicates the sum of the squares of the distances from each data point in the cluster to the cluster center; N is the total number of data to be clustered; the Calinski-Harabasz index under different numbers of centroids is compared. The larger the Calinski-Harabasz index, the tighter the cluster and the more dispersed the clusters are, that is, the better the clustering result. For example, when the number of centroids is 6, the Calinski-Harabasz index is the largest, and k groups of intelligent driving data are generated by clustering, and the k value is 6.

[0042] S403: Compare the k groups of data generated by clustering with the Q types of human driving styles obtained in step S300 and perform classification matching, and classify the k groups of data into the human driving styles respectively.

[0043] Clustering generates k groups of intelligent driving data, and calculates the median speed and median acceleration of each group of intelligent driving data through the target scene, and compares them with the median speed and median acceleration of Q types of human driving styles through the target scene. If the difference between this group of intelligent driving data and the driving data corresponding to one of the Q human driving styles is less than a preset value, the data is classified into that human driving style.

[0044] For example, the preset value is 15%. The median speed and acceleration of the first set of intelligent driving data passing through the target scene are 5.7m / s and 5.2m / s respectively. 2 In the steady human driving data, the median speed and acceleration of the vehicle passing through the target scene are 6m / s and 5m / s respectively. 2 , the gap between this group of intelligent driving data and the smooth human driving data is less than 15%, so the first group of intelligent driving data is classified as a smooth human driving style.

[0045] The method for calculating the gap in data comparison is: (|intelligent driving data - human driving data| / intelligent driving data)*100%. For example, the method for calculating the gap between the first set of intelligent driving data and the stable human driving data is that the difference in the median speed is [(6-5.7) / 5.7] *100%=5.3%; in addition, the difference in the median acceleration is [(5.2-5) / 5.2]*100%=3.8%. The gap between the first set of intelligent driving data and the stable human driving data is (5.3%, 3.8%).

[0046] In the present invention, when executing step S403: If the difference between any group of intelligent driving data in the k groups of data and two human driving style data in the Q types of human driving styles is smaller than a preset value, then the group of data is classified into the human driving style with the smaller difference; For example, the median speed and acceleration of the fifth set of intelligent driving data through the target scene were 6.6m / s and 5.6m / s respectively. 2 In the steady human driving data, the median speed and acceleration of the vehicle passing through the target scene are 6m / s and 5m / s respectively. 2 The median speed and acceleration of the vehicle passing through the target scene in the aggressive human driving data are 7m / s and 6m / s respectively. 2 , then the gap between the fifth group of intelligent driving data and the stable human driving style is (9.1%, 10.7%), and the gap between the fifth group of intelligent driving data and the stable human driving style is (6.1%, 7.1%). Although the gap between the fifth group of intelligent driving data and the two human driving style data is less than the preset value of 15%, the gap between it and the aggressive human driving style data is smaller, so the fifth group of intelligent driving data is classified as the aggressive human driving style; Here, each group of comparison gaps has two gap values, namely, the gap between the median speed and the median acceleration. When comparing two groups of comparison gaps, the comparison should be based on the average of the two gap values. For example, the gap between the sixth group of intelligent driving data and the stable human driving style is (5.5%, 6.4%), and the average of the two gap values ​​is 5.95%; the gap with the conservative human driving style is (5.7%, 6.1%), and the average of the two gap values ​​is 5.9%. The comparison gap between the sixth group of intelligent driving data and the two human driving style data is less than the preset value of 15%, but the comparison gap between it and the conservative human driving style data and the average of the comparison gap value are smaller. Therefore, the sixth group of intelligent driving data is classified into the conservative human driving style.

[0047] If the difference between any group of intelligent driving data in the k groups of data and the Q types of human driving style data is not less than the preset value, then the group of data is classified as other human driving styles; For example, the median speed and acceleration of the third set of intelligent driving data through the target scene are 3.3m / s and 1.6m / s respectively. 2 , when only three human driving styles are divided, the median speed and acceleration of the vehicle passing through the target scene in the conservative human driving data are 4m / s and 3m / s respectively 2 In the steady human driving data, the median speed and acceleration of the vehicle passing through the target scene are 6m / s and 5m / s respectively. 2 The median speed and acceleration of the vehicle passing through the target scene in the aggressive human driving data are 7m / s and 6m / s respectively. 2, then the difference between the third set of intelligent driving data and the three types of human driving style data is greater than 15% of the preset value, then the third set of intelligent driving data cannot be classified into any human driving style, and can only be classified into other human driving styles.

[0048] That is, after clustering Q groups of human driving data in step S200, there are actually Q+1 options when classifying the intelligent driving data, namely, Q human driving styles and other human driving styles.

[0049] It should be noted that the median speed and acceleration of a set of intelligent driving data passing through the target scene must be less than the preset value with the corresponding human driving data in order to be classified as the human driving style. For example For example, the median speed and acceleration of the second set of intelligent driving data through the target scene are 6.5m / s and 5.9m / s respectively. 2 In the steady human driving data, the median speed and acceleration of the vehicle passing through the target scene are 6m / s and 5m / s respectively. 2 , then the gap between the fifth set of intelligent driving data and the stable human driving style is (7.7%, 15.3%), among which the median comparison gap of speed is less than 15% of the preset value, but the median comparison gap of acceleration is greater than 15% of the preset value, so the second set of intelligent driving data cannot be classified into the stable human driving style.

[0050] S500: Calculate the anthropomorphic score of the intelligent driving vehicle according to the correction coefficient of the human driving style of the intelligent driving and the illegal situations of the intelligent driving during driving.

[0051] Before executing step S500 , a correction coefficient of the human driving style is assigned to each of the Q types of human driving style data and other human driving styles.

[0052] For example, when only three types of human driving styles are divided, namely conservative, smooth and aggressive, the correction coefficients assigned to the conservative human driving style are 0.9, the correction coefficients to the smooth human driving style are 0.8, the correction coefficients to the aggressive human driving style are 0.7, and the correction coefficients to other human driving styles are 0.5.

[0053] When executing step S500, specifically: S501: Obtaining a penalty point value L for illegal driving during driving; S502: Calculate the anthropomorphic score G of the intelligent driving vehicle according to the intelligent driving classification style; the anthropomorphic score calculation method is: G = (DL) * e * 100%; where D is the full score of the anthropomorphic evaluation; e is the correction coefficient of the human driving style corresponding to the driving data of the intelligent driving in driving; For example, the driving data of a certain intelligent driving vehicle in the target scenario is clustered into the intelligent driving data group, which is classified into the stable human driving style. The correction coefficient e of the stable human driving style is 0.8, the full score D of the anthropomorphic evaluation is 10 points, and the illegal deduction value L of the intelligent driving during driving is 2 points. Then the anthropomorphic score G of the intelligent driving vehicle is: (10-2)*0.8*100%, which is calculated to be 64 points.

[0054] Finally, it should be noted that when executing step S200 and step S402, the selected key parameters are clustered by the K-means++ algorithm to obtain multiple groups of data (the number of cluster centroids is k). The steps of the K-means++ algorithm are as follows: Step 1: In the data set X={x1,x2,...,x m} randomly selects a sample point x i As the initial centroid c1; Step 2: Calculate each remaining sample point x i The shortest distance d(x i ), calculate the probability p(x) of each sample point being selected as the next centroid i ),in: ; Step 3: Select the next centroid by roulette wheel method; Step 4: Repeat Step 2 and Step 3 until k centroids are selected, namely: c1, c2, ... c k ; Step 5: For each sample point x i Calculate x i The distance to k centroids, x i Divide into clusters with the centroids of the smallest distance; Step 6: Recalculate the centroid c of each cluster i ,in: ; Step 7: Repeat Step 5 and Step 6 until the center of mass no longer changes.

[0055] The present invention discloses an anthropomorphic evaluation method for intelligent driving vehicles. The driving data of the driver in a target scene is clustered and the human driving style is classified by using a principal component analysis method and a K-means++ algorithm. The intelligent human driving data is then clustered by using the K-means++ algorithm. The clustering result is matched to the human driving style. The final anthropomorphic score is obtained according to the classification of the human driving style and the illegal and irregular situations during driving. The anthropomorphic evaluation has high accuracy and standardization and is not easily interfered by subjective factors.

[0056] Embodiment 2: The embodiment of the present application provides an anthropomorphic evaluation system for intelligent driving vehicles, the system comprising: Control data feature acquisition module; Control data clustering module; Human driving style classification module; Intelligent driving style matching module; Anthropomorphic scoring module; The control data feature acquisition module obtains the characteristic parameters of human-driven vehicles in the target scene and selects key characteristic parameters; the control data clustering module clusters the selected key parameters to obtain Q groups of human driving data, where Q is not less than 3; the human driving style classification module calculates the median of the speed and the median of the acceleration of the vehicle passing through the target scene in the Q groups of human driving data, and corresponds the Q groups of human driving data to Q types of human driving styles; the intelligent driving style matching module clusters the data of the intelligent driving vehicle in the target scene, calculates the clustered multiple groups of data and compares them with the Q types of human driving styles for classification and matching; the classification and matching method is: calculate the median of the speed and the median of the acceleration of the clustered multiple groups of data passing through the target scene, and compares them with the median of the speed and the median of the acceleration of the vehicle passing through the target scene of any human driving style in the Q types of human driving styles. If the difference is less than the preset value, it is classified into the human driving style; the anthropomorphic scoring module calculates the anthropomorphic score of the intelligent driving vehicle according to the correction coefficient of the human driving style corresponding to the intelligent driving and the illegal deduction value of the intelligent driving during driving.

[0057] Those skilled in the art can clearly understand that the technical solutions of the embodiments of the present application can be implemented with the help of software and / or hardware. The "unit" and "module" in this specification refer to software and / or hardware that can independently complete or cooperate with other components to complete specific functions, where the hardware can be, for example, a field programmable gate array (FPGA), an integrated circuit (IC), etc.

[0058] It can be understood that the devices provided in the embodiments of the present invention are all applicable to the method described in Embodiment 1. The specific functions of each module can refer to the above method flow and will not be described in detail here.

[0059] Embodiment 3, an electronic device provided by the embodiment of the present invention is used to implement the method described in embodiment 1. The electronic device may include: at least one central processor, at least one network interface, a control interface, a memory, and at least one communication bus.

[0060] Among them, the communication bus is used to realize the connection, communication and information exchange between various components.

[0061] The network interface may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0062] Among them, the control interface is used to output control operations according to instructions.

[0063] The central processing unit may include one or more processing cores. The central processing unit uses various interfaces and lines to connect various parts in the entire terminal, and executes various functions of the terminal and processes data according to the method described in the first embodiment by running or executing instructions, programs, code sets or instruction sets stored in the memory, and calling data stored in the memory.

[0064] The memory may include a random access memory (RAM) or a read-only memory (ROM). Optionally, the memory includes a non-transitory computer-readable storage medium. The memory may be used to store instructions, programs, codes, code sets or instruction sets. The memory may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), and methods for implementing the above-mentioned embodiment 1; the data storage area may store data involved in each of the above method embodiments, etc.

[0065] The present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the method described in the first embodiment is implemented. The computer-readable storage medium may include, but is not limited to, any type of disk, including a floppy disk, an optical disk, a DVD, a CD-ROM, a micro drive, and a magneto-optical disk, a ROM, a RAM, an EPROM, an EEPROM, a DRAM, a VRAM, a flash memory device, a magnetic card or an optical card, a nanosystem (including a molecular memory IC), or any type of medium or device suitable for storing instructions and / or data.

[0066] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described order of actions, because according to the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0067] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0068] In the several embodiments provided by the present invention, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of the units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0069] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, the functional units in the various embodiments of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated units may be implemented in the form of hardware or in the form of software functional units.

[0070] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a memory, including several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned memory includes: U disk, read-only memory (ROM), random access memory (RAM), mobile hard disk, disk or optical disk, etc., various media that can store program codes.

[0071] A person skilled in the art may understand that all or part of the steps in the various methods of the above embodiments may be completed by entering a program to instruct related hardware, and the program may be stored in a computer-readable memory, which may include a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0072] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device comprising the elements.

[0073] The above contents are further detailed descriptions of the present invention in combination with specific preferred implementation modes, and it cannot be determined that the specific embodiments of the present invention are limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as belonging to the protection scope of the present invention.

Claims

1. A humanized evaluation method for intelligent driving vehicles, characterized in that: include: S100: Acquire characteristic parameters of a human driving a vehicle in a target scene, and select key characteristic parameters; S200: clustering the selected key parameters to obtain Q groups of human driving data; Where Q is not less than 3; S300: Calculate the median of the speed and the median of the acceleration of the vehicle passing through the target scene in the Q groups of human driving data, and correspond the Q groups of human driving data to Q types of human driving styles; S400: clustering the intelligent driving data of the intelligent driving vehicle in the target scenario, and comparing the clustered multiple groups of intelligent driving data with the Q types of human driving styles obtained in step S300 for classification and matching; The classification and matching method is: calculating the median of the speed and the median of the acceleration of the clustered multiple groups of intelligent driving data passing through the target scene, and comparing them with the median of the speed and the median of the acceleration of the vehicle passing through the target scene of any one of the Q human driving styles in step S300. If the difference is less than a preset value, it is classified into the human driving style; S500: Calculate the anthropomorphic score of the intelligent driving vehicle according to the correction coefficient of the human driving style corresponding to the intelligent driving and the penalty points for illegal driving during driving.

2. The method for anthropomorphic evaluation of an intelligent driving vehicle according to claim 1, characterized in that: When executing step S100, specifically: S101: Obtain m representative characteristic parameters of vehicle driving in a target scene; S102: reducing the dimensions of the m characteristic parameters by PCA principal component analysis, calculating the contribution rate of each characteristic parameter, and sorting them from large to small; S103: When the sum of the contribution rates of the first n characteristic parameters is not less than a preset threshold, the first n characteristic parameters are selected as key parameters.

3. The method for anthropomorphic evaluation of an intelligent driving vehicle according to claim 2, characterized in that: When executing step S101, the characteristic parameters include: average vehicle speed, maximum vehicle speed, standard deviation of vehicle speed, average vehicle lateral acceleration, maximum vehicle lateral acceleration, standard deviation of vehicle lateral acceleration, average vehicle longitudinal acceleration, maximum vehicle longitudinal acceleration, standard deviation of vehicle longitudinal acceleration, number of surrounding vehicles when passing through an intersection, and time to pass a target scene.

4. The method for anthropomorphic evaluation of an intelligent driving vehicle according to claim 1, characterized in that: When executing step S200, specifically: the number of centroids is set to Q, the selected key parameters are clustered by the K-means++ algorithm, and finally Q groups of human driving data are obtained.

5. The method for anthropomorphic evaluation of an intelligent driving vehicle according to claim 1, characterized in that: When executing step S300, the median of the speed and the median of the acceleration of the vehicle passing through the target scene in the Q group of data are calculated and compared, and the Q group of human driving data are arranged in order from small to large, and the Q group of human driving data are corresponded to Q types of human driving styles, and the Q group of data are classified into human driving styles; Among them: the lower median speed and median acceleration are conservative human driving styles, the medium median speed and median acceleration are stable human driving styles, and the larger median speed and median acceleration are aggressive human driving styles.

6. The method for anthropomorphic evaluation of an intelligent driving vehicle according to claim 5, characterized in that: When executing step S400, specifically: S401: Acquire driving data of an intelligent driving vehicle in a target scenario as data to be clustered; the driving data includes m characteristic parameters and n key parameters of the intelligent driving vehicle in the target scenario; S402: clustering the data to be clustered using the K-means++ algorithm to generate k groups of data; The number of centroids ranges from 2 to 9. The Calinski-Harabasz index under different numbers of centroids is calculated, and the number of centroids k when the Calinski-Harabasz index is the largest is selected. S403: Compare the k groups of data generated by clustering with the Q types of human driving styles obtained in step S300 and perform classification matching, and classify the k groups of data into the human driving styles respectively.

7. The method for anthropomorphic evaluation of an intelligent driving vehicle according to claim 6, characterized in that: When executing step S403, If the difference between any group of intelligent driving data in the k groups of data and two human driving style data in the Q types of human driving styles is smaller than a preset value, then the group of data is classified into the human driving style with the smaller difference; If the difference between any group of intelligent driving data in the k groups of data and the Q types of human driving style data is not less than the preset value, then the group of data will be classified as other human driving styles.

8. The method for anthropomorphic evaluation of an intelligent driving vehicle according to claim 7, characterized in that: Before executing step S500 , a correction coefficient of the human driving style is assigned to each of the Q types of human driving style data and other human driving styles.

9. The method for anthropomorphic evaluation of an intelligent driving vehicle according to claim 8, characterized in that: When executing step S500, specifically: S501: Obtaining a penalty point value L for illegal driving during driving; S502: Calculate the anthropomorphic score G of the intelligent driving vehicle according to the intelligent driving classification style; the anthropomorphic score calculation method is: G=(DL)*e*100%; where D is the full score of the anthropomorphic evaluation; e is the correction coefficient of the human driving style corresponding to the intelligent driving during driving.

10. An anthropomorphic evaluation system for intelligent driving vehicles, characterized in that: include: Control data feature acquisition module; Control data clustering module; Human driving style classification module; Intelligent driving style matching module; Anthropomorphic scoring module; The reference data feature acquisition module obtains the characteristic parameters of human driving vehicles in the target scene and selects key characteristic parameters; the reference data clustering module clusters the selected key parameters to obtain Q groups of human driving data, where Q is not less than 3; the human driving style classification module calculates the median of the speed and the median of the acceleration of the vehicle passing through the target scene in the Q groups of human driving data, and corresponds the Q groups of human driving data to Q types of human driving styles; The intelligent driving style matching module clusters the data of the intelligent driving vehicle in the target scene, calculates the multiple sets of clustered data and compares them with Q types of human driving styles for classification and matching; the classification and matching method is: calculate the median of the speed and the median of the acceleration of the multiple sets of clustered data passing through the target scene, and compare them with the median of the speed and the median of the acceleration of the vehicle passing through the target scene with any of the Q types of human driving styles. If the difference is less than a preset value, it is classified into the human driving style; The anthropomorphic scoring module calculates the anthropomorphic score of the intelligent driving vehicle based on the correction coefficient of the human driving style corresponding to the intelligent driving and the illegal deduction points for the intelligent driving during driving.

Images