Vehicle data testing method, system, medium, and device based on comfort evaluation

Abstract

The application provides a vehicle data test method and system based on comfort evaluation, a medium and equipment, comprising the following steps: S1, establishing subjective evaluation indexes and objective evaluation indexes of a vehicle, obtaining subjective scores and objective scores of comfort evaluation respectively, and obtaining final comfort scores through weighted normalization; S2, building a comfort evaluation model through collected vehicle motion data and final comfort scores; S3, classifying and extracting non-comfort scene data according to the comfort evaluation model; and S4, using the classified scene data for automatic driving algorithm improvement and new automatic driving algorithm feasibility verification. The application discloses a method for scoring comfort of an automatic driving vehicle, subjective and objective evaluation indexes are proposed, and the final comfort score of the automatic driving vehicle is obtained through weighting of the subjective indexes and the objective indexes.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of automatic driving, in particular, to a vehicle data test method and system based on comfort evaluation, a medium and equipment, more particularly, to an automatic driving vehicle data closed-loop test method and system based on comfort evaluation, a medium and equipment. BACKGROUND

[0002] With the rapid development of automatic driving technology, in addition to ensuring the safety of automatic driving, the comfort of automatic driving vehicles is increasingly concerned by people, and how to continuously improve the comfort of automatic driving vehicles has become a key point in the development of automatic driving technology. Moreover, the evaluation of comfort has individual subjective differences, and subjective feelings alone cannot give an effective and accurate judgment of the comfort of automatic driving vehicles, so a method for evaluating the comfort of automatic driving vehicles is needed. At the same time, automatic driving vehicles collect a large amount of automatic driving data every day, and if there is no efficient data extraction method, the huge recorded data cannot help technology developers quickly find problems and then iterate and upgrade algorithms, so how to efficiently use data and not let repetitive data occupy local disks or servers has become a problem to be solved.

[0003] How to continuously improve the comfort of automatic driving vehicles has become a key point in the development of automatic driving technology. A large amount of original data generated by automatic driving vehicles every day, how to quickly extract non-comfort scene data from them, if only relying on the problem records and feedbacks of test personnel, because of individual differences, some problems may be ignored, and if a large amount of recorded data is not marked and classified, it will waste disk or server resources, and this resource is not used reasonably and efficiently.

[0004] The invention document CN112353393B discloses a method for evaluating the comfort of passengers in an intelligent driving car, comprising: 1) obtaining experimental data, including the physical information of each measured passenger in a static state and an intelligent driving car running state, and the subjective comfort evaluation index of the measured passenger; 2) based on the obtained physical information and the pre-established comfort objective evaluation model based on physical information, calculating the comfort objective evaluation index based on physical information; 3) based on the dynamic index and the passenger comfort prediction model based on vehicle dynamics, obtaining the passenger comfort prediction evaluation index based on vehicle dynamics information; 4) constructing a passenger comfort comprehensive evaluation model, based on the passenger comfort comprehensive evaluation index predicted by the passenger comfort comprehensive evaluation model and the vehicle three-degree-of-freedom model, establishing a vehicle dynamics control domain based on passenger comfort to ensure ride comfort.

[0005] The application establishes a method for evaluating the comfort of passengers in intelligent driving cars by correlating the collected electromyography, electrocardiography and electroencephalography signals of passengers with their subjective comfort ratings, and explores the relationship between the dynamic control of intelligent driving cars such as steering, braking and acceleration and the subjective comfort of passengers. The main shortcomings of the application are: 1) different age groups and different genders are not sampled because their body signals are different; 2) it is difficult to collect data during the test, the cost is high, and the repeatability is poor, and the consistency of data collection is poor due to changes in the environment; 3) only the comfort evaluation model is obtained, and how to feed back the comfort evaluation results / data to the intelligent driving system to improve the system performance is not mentioned.

[0006] Patent document CN109177979B discloses a data processing method and device for evaluating the comfort of a car ride and a readable storage medium. It discloses a method for evaluating the comfort of a car ride by receiving evaluation data input by a user through a data collection port, the evaluation data including evaluation information provided by the user on each driving action of the vehicle he / she is riding, determining environmental information and / or vehicle driving parameters when the vehicle performs each driving action, and training a preset deep learning algorithm model according to the evaluation information corresponding to each driving action of the vehicle, the environmental information and / or the vehicle driving parameters, to obtain an evaluation model for outputting the comfort of a car ride. However, this invention has the following shortcomings: 1. The comfort of a car ride is evaluated only by the subjective ratings of the experimenter, which is too single, inaccurate and subjective, and not objective; 2. The invention does not mention the objective indicators related to comfort evaluation and the calculation method of comfort ratings; 3. The invention does not involve the automatic extraction of non-comfort scenario data and the use of the extracted data.

[0007] Patent document CN114841514A discloses a model training and vehicle comfort evaluation method, device, equipment and storage medium. The method includes: obtaining sample vehicle driving data and sample comfort evaluation data corresponding to the sample vehicle driving data; the sample comfort evaluation data is the evaluation data given by a user on the comfort of a vehicle after driving the vehicle based on the sample vehicle driving data; determining the sample vehicle body frequency response characteristics according to the sample vehicle driving data; training a neural network model using the sample vehicle body frequency response characteristics and the sample comfort evaluation data to obtain a comfort evaluation model, which is used to evaluate the comfort of a vehicle. However, this invention does not involve the automatic extraction of non-comfort scenario data and the use of the extracted data.

[0008] The patent document CN111667605B discloses an automatic driving test data storage method, device and electronic equipment, and relates to the technical field of automatic driving. The method comprises the following steps: obtaining a task type of an automatic driving vehicle test task; receiving message data corresponding to the task type in the automatic driving vehicle test, and writing the message data into a data queue; creating a record file corresponding to the task type, writing the message data corresponding to the task type in the data queue into the record file, and transmitting the record file and the file name of the record file to the data queue; obtaining preset disk landing scene information corresponding to the task type of the automatic driving vehicle test task, wherein the preset disk landing scene information represents a time of triggering data disk landing storage; and adding an unerasable mark to a record file in the data queue, wherein the data acquisition time of the record file is consistent with the time of triggering data disk landing storage, according to the preset disk landing scene information corresponding to the task type of the automatic driving vehicle test task. However, the invention does not mention the objective indicators related to comfort evaluation and the calculation method of comfort score.

[0009] The patent document CN111858927A discloses a data test method, device, electronic equipment and storage medium, and relates to the field of automatic driving. The method can comprise the following steps: obtaining a first data set, running each piece of data in the first data set, and judging each piece of data according to the running result to obtain the judgment result of whether each piece of data passes; determining the scene classification to which each piece of data in the first data set belongs; mapping each piece of data in each scene classification to a statistical graph corresponding to the scene classification, which is generated in advance, according to the mapping result and the judgment result of each piece of data in the scene classification, and generating a test evaluation index corresponding to the scene classification. However, the invention does not mention the objective indicators related to comfort evaluation and the calculation method of comfort score. SUMMARY

[0010] In view of the defects in the prior art, the purpose of the present application is to provide a vehicle data test method, system, medium and equipment based on comfort evaluation.

[0011] According to the vehicle data test method based on comfort evaluation provided by the present application, the following steps are included:

[0012] Step S1: Establishing subjective evaluation indicators and objective evaluation indicators of the vehicle, obtaining subjective scores and objective scores of comfort evaluation respectively, and obtaining a final comfort score through weighted normalization;

[0013] Step S2: Building a comfort evaluation model through the collected vehicle motion data and the final comfort score;

[0014] Step S3: Classifying and extracting scene data of discomfort according to the comfort evaluation model based on the vehicle motion data;

[0015] Step S4: using the classified scene data for automatic driving algorithm improvement and new automatic driving algorithm feasibility verification.

[0016] Preferably, in the step S1:

[0017] Based on the ISO 2631 international standard, the comfort score of the vehicle is established;

[0018] The test scene for comfort evaluation is built, including the automatic driving road test scene and the scene during the actual driving of the vehicle;

[0019] Data collection: including objective data and subjective data:

[0020] The objective data includes the vehicle driving parameters in the ISO 2631 standard, including the acceleration of the vehicle in the lateral, longitudinal and vertical directions, and the vibration frequency corresponding to the three accelerations, the speed of the vehicle, the steering wheel angle, the angular velocity of the vehicle and the environmental temperature during the operation of the vehicle;

[0021] The subjective data includes the comfort evaluation scores given by people of different ages and different genders after riding the automatic driving vehicle;

[0022] The evaluation index includes objective evaluation index and subjective evaluation index;

[0023] The objective evaluation index includes acceleration index, car sickness value and objective comfort value;

[0024] The acceleration index includes:

[0025] Effective weighted root mean square acceleration:

[0026]

[0027] Wherein, a w represents the effective weighted root mean square acceleration, T is the measurement period, a w (t) is the real-time equivalent acceleration, wherein is the original lateral, longitudinal and vertical acceleration collected by the vehicle, h is the change frequency of the acceleration, is a function of the variable h;

[0028] Equivalent acceleration:

[0029]

[0030] Wherein, w i is the weight coefficient, represents the lateral, longitudinal and vertical effective weighted root mean square acceleration, respectively;

[0031] Car sickness value:

[0032]

[0033] T is the total time of vibration occurrence;

[0034] Objective comfort value a eq :

[0035] a eq = f(a, MSDVz)

[0036] The subjective evaluation index includes the subjective evaluation comfort score of the feeling group of different ages and different genders;

[0037] The objective comfort score a mentioned in the evaluation index eq And the subjective evaluation comfort score is normalized according to different weighting coefficients to obtain the final comfort score of the vehicle.

[0038] Preferably, in the step S2:

[0039] A vehicle comfort evaluation model is built through a neural network;

[0040] According to the comfort score and the motion data of the vehicle, a vehicle comfort evaluation model is built through a neural network using a preset proportion of original data;

[0041] The remaining original data is used to verify the comfort evaluation model;

[0042] The verified comfort evaluation model is used for comfort evaluation of the vehicle.

[0043] Preferably, in the step S3:

[0044] Scene data below the comfort score threshold in the automatic driving process is classified and extracted, which is used for algorithm analysis and iteration of the automatic driving system;

[0045] The comfort evaluation model is run on the vehicle in real time to extract non-comfort scene data, which is analyzed by reading the recorded data packet form to obtain non-comfort scene data;

[0046] The non-comfort scene data is respectively given different data labels according to the reasons for exceeding the comfort evaluation threshold;

[0047] The classified scene data is used as a data basis for automatic driving algorithm analysis, algorithm logic and parameter adjustment;

[0048] The adjusted automatic driving algorithm is verified using the extracted scene data to realize the test closed loop of the classified scene data.

[0049] The application provides a vehicle data test system based on comfort evaluation, which executes the vehicle data test method based on comfort evaluation, and comprises the following steps of:

[0050] A data collection module is arranged to collect original data of each module of the vehicle.

[0051] A comfort evaluation module is arranged to build a comfort evaluation model and apply the evaluation model to the comfort evaluation process of the vehicle in real time.

[0052] A scene data extraction module is arranged to classify and extract scene data of non-comfort according to the comfort evaluation model and the vehicle motion data.

[0053] A scene data classification module is arranged to extract and classify the scene data of non-comfort, and mark the scene data with different scene labels according to the reasons of the non-comfort score.

[0054] An algorithm analysis module is arranged to perform logical verification and parameter adjustment of the driving algorithm and feasibility verification of the new algorithm according to the classified scene data processed by the scene classification module, so as to realize the closed loop of the test data.

[0055] A scene data storage module is arranged to store the classified scene data, so as to facilitate the algorithm analysis and the feasibility verification of the new algorithm.

[0056] Preferably, the comfort score of the vehicle is established based on the ISO 2631 international standard.

[0057] The test scene of the comfort evaluation is built, including the automatic driving road test scene and the scene during the actual driving of the vehicle.

[0058] The data collection includes objective data and subjective data.

[0059] The objective data includes the vehicle driving parameters in the ISO 2631 standard, including the acceleration of the vehicle in the lateral direction, the longitudinal direction and the vertical direction, the vibration frequency corresponding to the three accelerations, the speed of the vehicle, the steering wheel angle, the angular velocity of the vehicle and the environmental temperature during the operation of the vehicle.

[0060] The subjective data includes the comfort evaluation scores given by people of different ages and different genders after riding the automatic driving vehicle.

[0061] The evaluation indexes include objective evaluation indexes and subjective evaluation indexes.

[0062] The objective evaluation indexes include the acceleration index, the car sickness value and the objective comfort value.

[0063] The acceleration index includes:

[0064] The effective weighted root mean square acceleration:

[0065]

[0066] wherein a w represents the effective weighted root mean square acceleration, T is the measurement period, a w (t) is the real-time equivalent acceleration, wherein is the original lateral, longitudinal and vertical acceleration collected by the vehicle, h is the frequency of acceleration change, is a variable function of h;

[0067] Equivalent acceleration:

[0068]

[0069] wherein w i is a weight coefficient, respectively represents the lateral, longitudinal and vertical effective weighted root mean square acceleration;

[0070] Car sickness value:

[0071]

[0072] wherein T is the entire time of vibration occurrence;

[0073] Objective comfort value a eq :

[0074] a eq =f(a,MSDVz)

[0075] The subjective evaluation index includes the subjective evaluation comfort score of the feeling group of different ages and different genders;

[0076] The objective comfort score a eq mentioned in the evaluation index and the subjective evaluation comfort score are normalized according to different weight coefficients to obtain the final comfort score of the vehicle.

[0077] Preferably, the vehicle comfort evaluation model is built by a neural network;

[0078] According to the comfort score and the motion data of the vehicle, the original data of a preset proportion are used to build the vehicle comfort evaluation model by a neural network;

[0079] The remaining original data are used to verify the comfort evaluation model;

[0080] The verified comfort evaluation model is used for the comfort evaluation of the vehicle.

[0081] Preferably, the scene data below the comfort type score threshold in the automatic driving process are classified and extracted for algorithm analysis and iteration of the automatic driving system.

[0082] The comfort evaluation model is run on the vehicle in real time to extract non-comfort scene data, and the non-comfort scene data is obtained by reading and analyzing recorded data packets;

[0083] The non-comfort scene data is respectively given different data labels according to reasons for exceeding the comfort evaluation threshold;

[0084] The classified scene data is used as a data basis for automatic driving algorithm analysis, algorithm logic and parameter adjustment;

[0085] The adjusted automatic driving algorithm is verified by using the extracted scene data, so that a test closed loop of the classified scene data is realized.

[0086] According to the present application, a computer readable storage medium storing a computer program is provided, and the computer program is executed by a processor to realize the steps of the vehicle data test method based on comfort evaluation.

[0087] According to the present application, a vehicle device based on comfort evaluation is provided, which comprises a controller.

[0088] The controller comprises the computer readable storage medium storing the computer program, and the computer program is executed by a processor to realize the steps of the vehicle data test method based on comfort evaluation; or the controller comprises the vehicle data test system based on comfort evaluation.

[0089] Compared with the prior art, the present application has the following beneficial effects:

[0090] 1. The present application discloses a method for evaluating the comfort of an automatic driving vehicle, and subjective and objective evaluation indexes are proposed, and the final comfort score of the automatic driving vehicle is obtained by weighting the subjective and objective indexes.

[0091] 2. The present application uses a neural network to build a comfort evaluation model for the automatic driving vehicle, and through data learning in the early stage, more accurate and personified comfort evaluation can be directly given according to the dynamic performance of the vehicle in real time.

[0092] 3. The non-comfort scene in the automatic driving process of the vehicle is directly extracted and classified into different labeled scene data sets, which are used for improving the automatic driving algorithm and verifying the feasibility of the new algorithm.

[0093] 4. The present application only records the non-comfort scene data, but not all automatic driving data, thereby saving storage resources. BRIEF DESCRIPTION OF DRAWINGS

[0094] Other features, objects, and advantages of the application will become apparent from the following detailed description of non-limiting embodiments, when read in connection with the following accompanying drawings:

[0095] Figure 1 A schematic diagram of a system flow of the application;

[0096] Figure 2 A schematic diagram of a comfort evaluation model for an autonomous vehicle;

[0097] Figure 3 A schematic diagram of a non-comfort scene data classification automation extraction and algorithm analysis flow;

[0098] Figure 4 A schematic diagram of an autonomous driving evaluation device. DETAILED DESCRIPTION

[0099] The application will be described in detail below with specific embodiments. The following embodiments will help those skilled in the art to further understand the application, but do not limit the application in any form. It should be noted that for those skilled in the art, without departing from the concept of the application, a number of changes and improvements can be made. These are within the scope of the application.

[0100] Example 1:

[0101] The application proposes to establish a comfort evaluation method for autonomous vehicles, and to build a comfort evaluation model for autonomous vehicles. The verified model is run in real time on the data package recorded by the autonomous vehicle, and the scene data below the comfort threshold is classified and extracted for the verification of the logic / parameters of the autonomous driving algorithm. At the same time, it can also be used as scene data for the feasibility verification of new algorithms to realize the closed loop of test data. The comfort of autonomous vehicles is continuously improved by using test data.

[0102] According to the vehicle data test method based on comfort evaluation provided by the application, as shown in Figures 1-4 , comprising:

[0103] Step S1: Establishing subjective evaluation indexes and objective evaluation indexes of the vehicle, obtaining subjective scores and objective scores of the comfort evaluation respectively, and obtaining final comfort scores through weighted normalization;

[0104] Specifically, in the step S1:

[0105] Based on the ISO 2631 international standard, the comfort scores of the vehicle are established;

[0106] The test scene of the comfort evaluation is built, including the autonomous driving road test scene and the scene during the actual driving of the vehicle;

[0107] Data collection: including objective data and subjective data:

[0108] Objective data includes vehicle driving parameters in ISO 2631 standard, including vehicle acceleration in lateral, longitudinal and vertical directions, and vibration frequency corresponding to the three accelerations respectively, vehicle speed, steering wheel angle, vehicle angular velocity and environmental temperature of vehicle operation;

[0109] Subjective data includes comfort evaluation scores given by people of different ages and different genders after riding an automatic driving vehicle;

[0110] Evaluation index includes objective evaluation index and subjective evaluation index;

[0111] Objective evaluation index includes acceleration index, car sickness value and objective comfort value;

[0112] Acceleration index includes:

[0113] Effective weighted root mean square acceleration:

[0114]

[0115] Wherein, a w represents effective weighted root mean square acceleration, T is the measurement period, a w (t) is real-time equivalent acceleration, wherein is the original lateral, longitudinal and vertical acceleration collected by the vehicle, h is the change frequency of acceleration, is a variable function of h;

[0116] Equivalent acceleration:

[0117]

[0118] Wherein, w i is a weight coefficient, represents lateral, longitudinal and vertical effective weighted root mean square acceleration respectively;

[0119] Car sickness value:

[0120]

[0121] Wherein, T is the whole time of vibration occurrence;

[0122] Objective comfort value a eq :

[0123] a eq =f(a,MSDVz)

[0124] The subjective evaluation index includes subjective evaluation comfort scores of different age groups and different gender groups;

[0125] The objective comfort score a mentioned in the evaluation index eq And the subjective evaluation comfort score is normalized according to different weighting coefficients to obtain the final comfort score of the vehicle.

[0126] Step S2: building a comfort evaluation model through the collected vehicle motion data and the final comfort score;

[0127] Specifically, in the step S2:

[0128] The vehicle comfort evaluation model is built through the neural network;

[0129] According to the comfort score and the motion data of the vehicle, the vehicle comfort evaluation model is built through the neural network using the original data in a preset proportion;

[0130] The remaining original data is used to verify the comfort evaluation model;

[0131] The verified comfort evaluation model is used for comfort evaluation of the vehicle.

[0132] Step S3: classifying and extracting non-comfort scene data according to the comfort evaluation model based on the vehicle motion data;

[0133] Specifically, in the step S3:

[0134] The scene data below the comfort score threshold in the automatic driving process is classified and extracted, which is used for algorithm analysis and iteration of the automatic driving system;

[0135] The comfort evaluation model is run on the vehicle in real time to extract non-comfort scene data, which is analyzed by reading the recorded data packets to obtain non-comfort scene data;

[0136] The non-comfort scene data is respectively given different data labels according to the reasons for exceeding the comfort evaluation threshold;

[0137] The classified scene data is used as a data basis for automatic driving algorithm analysis, algorithm logic and parameter adjustment;

[0138] The adjusted automatic driving algorithm is verified using the extracted scene data to realize a test closed loop of the classified scene data.

[0139] Step S4: using the classified scene data for automatic driving algorithm improvement and new automatic driving algorithm feasibility verification.

[0140] Embodiment 2:

[0141] Example 2 is a preferred example of Example 1, to more specifically illustrate the present application.

[0142] The present application also provides a comfort evaluation based vehicle data testing system, which can be implemented by performing the process steps of the comfort evaluation based vehicle data testing method, i.e. the comfort evaluation based vehicle data testing method can be understood by those skilled in the art as a preferred embodiment of the comfort evaluation based vehicle data testing system.

[0143] According to the present application, a comfort evaluation based vehicle data testing system is provided, which performs the comfort evaluation based vehicle data testing method, including:

[0144] Data acquisition module: collecting original data of each module of the vehicle;

[0145] Comfort evaluation module: building a comfort evaluation model and applying the evaluation model in real time in the comfort evaluation process of the vehicle;

[0146] Scene data extraction module: classifying and extracting non-comfort scene data according to the comfort evaluation model based on the vehicle motion data;

[0147] Scene data classification module: extracting non-comfort scene data classification, and marking different scene labels on the scene data according to the reasons for non-comfort scores;

[0148] Algorithm analysis module: performing logical verification and parameter adjustment of driving algorithms and feasibility verification of new algorithms based on the classified scene data processed by the scene classification module, to realize closed loop of test data;

[0149] Scene data storage module: storing the classified scene data for algorithm analysis and feasibility verification of new algorithms.

[0150] Specifically, based on the ISO 2631 international standard, the comfort score of the vehicle is established;

[0151] Building a comfort evaluation test scene, including an automatic driving road test scene and a scene during actual vehicle driving;

[0152] Data acquisition: including objective data and subjective data:

[0153] Objective data includes vehicle driving parameters in the ISO 2631 standard, including vehicle acceleration in lateral, longitudinal and vertical directions, vibration frequency corresponding to the three accelerations, vehicle speed, steering wheel angle, vehicle angular velocity and environmental temperature during vehicle operation;

[0154] The subjective data include comfort evaluation scores given by people of different ages and different genders after they ride the autonomous vehicle;

[0155] The evaluation indexes include objective evaluation indexes and subjective evaluation indexes;

[0156] The objective evaluation indexes include acceleration indexes, car sickness values and objective comfort values;

[0157] The acceleration indexes include:

[0158] The effective weighted root mean square acceleration:

[0159]

[0160] wherein a w represents the effective weighted root mean square acceleration, T is a measurement period, a w (t) is a real-time equivalent acceleration, wherein is the original lateral, longitudinal and vertical acceleration collected by the vehicle, h is the change frequency of the acceleration, is a variable function of h;

[0161] The equivalent acceleration:

[0162]

[0163] wherein w i is a weight coefficient, respectively represent the lateral, longitudinal and vertical effective weighted root mean square acceleration;

[0164] The car sickness value:

[0165]

[0166] wherein T is the entire time of vibration occurrence;

[0167] The objective comfort value a eq :

[0168] a eq =f(a,MSDVz)

[0169] The subjective evaluation indexes include subjective evaluation comfort scores of the feeling groups of different ages and different genders;

[0170] The objective comfort scores a eq and the subjective evaluation comfort scores mentioned in the evaluation indexes are normalized according to different weight coefficients to obtain the final comfort scores of the vehicle.

[0171] Specifically, a vehicle comfort evaluation model is built through a neural network;

[0172] According to the comfort score and the motion data of the vehicle, a vehicle comfort evaluation model is built by a neural network using preset proportions of the original data;

[0173] The remaining original data is used to verify the comfort evaluation model;

[0174] The verified comfort evaluation model is used for comfort evaluation of the vehicle.

[0175] Specifically, scene data below a comfort type score threshold in the automatic driving process is classified and extracted for algorithm analysis and iteration of the automatic driving system;

[0176] The comfort evaluation model is used to extract non-comfort scene data in real time on the vehicle, and the non-comfort scene data is obtained by reading and analyzing the recorded data packets;

[0177] The non-comfort scene data is respectively given different data labels according to reasons for exceeding the comfort evaluation threshold;

[0178] The classified scene data is used as a data basis for algorithm analysis, algorithm logic and parameter adjustment of the automatic driving system;

[0179] The adjusted automatic driving algorithm is verified using the extracted scene data to realize a test closed loop of the classified scene data.

[0180] According to the present application, a computer readable storage medium storing a computer program is provided, and the computer program is executed by a processor to implement the steps of the vehicle data test method based on comfort evaluation.

[0181] According to the present application, a vehicle device based on comfort evaluation is provided, which comprises a controller.

[0182] The controller comprises the computer readable storage medium storing the computer program, and the computer program is executed by a processor to implement the steps of the vehicle data test method based on comfort evaluation; or the controller comprises the vehicle data test system based on comfort evaluation.

[0183] Embodiment 3:

[0184] Embodiment 3 is a preferred example of Embodiment 1, which is used to more specifically illustrate the present application.

[0185] Overall introduction of the system scheme:

[0186] 1. By establishing subjective evaluation indexes and objective evaluation indexes of the automatic driving vehicle, subjective scores and objective scores of comfort evaluation are obtained, respectively;

[0187] 2. Using subjective and objective scores, the final comfort score is obtained by weighted normalization;

[0188] 3. Using the collected motion data of autonomous vehicles and the final comfort score, an autonomous driving comfort evaluation model is built using neural networks;

[0189] 4. The real-time running data or collected historical data of autonomous vehicles are automatically classified and extracted according to the comfort evaluation model to obtain non-comfort scene data;

[0190] 5. The classified scene data is used for algorithm improvement and new algorithm feasibility verification;

[0191] Among them:

[0192] 1. Based on the ISO 2631 (Mechanical vibration and shock - Evaluation of human exposure to whole-body vibration) international standard, the comfort score of autonomous vehicles is established;

[0193] a. Test scene

[0194] The test scene for comfort evaluation is built, including but not limited to the scene when the vehicle is manually driven in daily life;

[0195] b. Data collection

[0196] i. Objective data: vehicle driving parameters mentioned in the ISO 2631 standard, including but not limited to vehicle acceleration in lateral / longitudinal / vertical direction, vibration frequency corresponding to three accelerations, vehicle speed, steering wheel angle, vehicle angular velocity and vehicle operating environment temperature;

[0197] ii. Subjective data: collect comfort evaluation scores given by people of different ages and different genders after experiencing autonomous vehicles;

[0198] c. Evaluation index

[0199] i. Objective evaluation index

[0200] 1) Acceleration index

[0201] Effective weighted root mean square acceleration:

[0202] Where a w represents the effective weighted root mean square acceleration, T is the measurement period, a w (t) represents the real-time equivalent acceleration, where Where is the original lateral / longitudinal / vertical acceleration collected by the autonomous vehicle, h is the change frequency of the acceleration, is the variable h is a function of;

[0203] Equivalent acceleration where w i is a weight coefficient,

[0204] respectively represent the lateral / longitudinal / vertical effective weighted root mean square acceleration;

[0205] 2) Motion sickness dose value (MSDVz):

[0206] where T refers to the entire time of vibration occurrence; a w (t) represents the real-time equivalent acceleration in the z direction.

[0207] 3) Objective comfort value / a eq : Objective comfort score established by a function relationship between the equivalent acceleration / a and the motion sickness value / MSDVz

[0208] a eq = f(a, MSDVz)

[0209] ii. Subjective evaluation index

[0210] Subjective evaluation comfort score of the feeling group of different ages and different genders;

[0211] d. Final score

[0212] Objective comfort score a mentioned in the evaluation index eq and subjective evaluation comfort score are normalized according to different weight coefficients to obtain the final comfort score of the autonomous vehicle;

[0213] 2. Build a model of the comfort score of the autonomous vehicle and the vehicle dynamics performance through a neural network;

[0214] a. According to the mentioned comfort score and vehicle dynamics performance data, use 80% of the original data to build a model of the comfort score of the autonomous vehicle and the vehicle dynamics performance through a neural network;

[0215] b. Use the remaining 20% of the original data to verify the comfort evaluation model;

[0216] c. Use the verified comfort evaluation model for comfort evaluation of the autonomous vehicle;

[0217] Comfort evaluation model building process:

[0218] Process introduction:

[0219] a. Through the mentioned subjective evaluation index and objective evaluation index, respectively obtain subjective score and objective score, and obtain final comfort score under different test scenes through weighted normalization calculation;

[0220] b. Based on the collected 80% data set (vehicle motion parameters and final comfort score), the corresponding relationship between vehicle motion parameters and autonomous vehicle comfort comfort score is established by using neural network, and then the comfort evaluation model of autonomous vehicle is obtained;

[0221] c. The remaining 20% data set is used for comfort model verification, which is used for subsequent automatic comfort evaluation of autonomous driving;

[0222] Among them, the subjective score suggestion (for reference):

[0223] Comfort level Comfort score Comfort 10 Slightly uncomfortable 8 Somewhat uncomfortable 6 Uncomfortable 4 Very uncomfortable 2 Extremely uncomfortable 0

[0224] Among them, the objective score principle (for reference):

[0225]

[0226]

[0227] 3. The test data of autonomous driving is classified and automatically extracted from the scene data below the comfort score threshold in the process of autonomous driving, which is used for algorithm analysis and algorithm iteration of autonomous driving system;

[0228] a. The comfort evaluation function can be run in real time on the autonomous vehicle to extract non-comfort scene data, or the recorded data packet can be analyzed to obtain non-comfort scene data;

[0229] b. Non-comfort scene data is respectively given different data labels according to the reason of exceeding the comfort evaluation threshold;

[0230] c. The classified scene data is used as data basis for autonomous driving algorithm analysis, algorithm logic and parameter adjustment;

[0231] d. The adjusted algorithm / new algorithm is verified by using the extracted scene data of each type, realizing the test closed loop of classified scene data;

[0232] An apparatus of autonomous vehicle based on comfort evaluation, comprising:

[0233] a. Data acquisition module, used for collecting original data of each module of autonomous vehicle;

[0234] b. Comfort evaluation module, used for building comfort evaluation model of vehicle, and applying the evaluation model in real time in the comfort evaluation process of autonomous vehicle;

[0235] c. A scene data extraction module for extracting scene data that does not meet the comfort threshold according to the comfort score given by the comfort evaluation module during the autonomous driving process and / or the autonomous driving data playback process;

[0236] d. A scene data classification module for classifying the extracted non-comfort scene data, and labeling the scene data with different scene labels according to the reasons for the non-comfort score, for the algorithm verification module;

[0237] e. An algorithm analysis module for logical verification, parameter adjustment, and feasibility verification of new algorithms of the autonomous driving algorithm based on the classified scene data processed by the scene classification module, to realize the closed loop of test data.

[0238] f. A scene data storage module mainly for storing the classified scene data to facilitate algorithm analysis and feasibility verification of new algorithms;

[0239] An electronic device of an autonomous driving vehicle based on comfort evaluation, comprising:

[0240] a. One or more processors;

[0241] b. A storage device;

[0242] i. For storing one or more programs;

[0243] ii. For storing the collected original data of the autonomous driving vehicle and the classified scene data;

[0244] 1) The original data of the autonomous driving vehicle can be stored according to the storage capacity of the hard disk / server and the data demand to evaluate whether to store;

[0245] 2) The classified scene data is small in data volume and clear in problem, so that the data storage is convenient for verification of the changed algorithm and feasibility verification of the new algorithm;

[0246] An electronic device of an autonomous driving vehicle based on comfort evaluation, comprising:

[0247] The data storage medium is a readable storage medium, and a computer program is stored on the data storage medium, wherein the program is executed by a processor to realize the functions of the present application.

[0248] Among them, based on the comfort evaluation index mentioned in the present application, data collection, subjective and objective index weight coefficient and part of the index can be changed according to the characteristics of the autonomous driving vehicle;

[0249] The comfort evaluation model of the autonomous driving vehicle can also be established by using reinforcement learning or deep learning method, which is not limited to neural network model;

[0250] The extraction and classification method of scene data is not limited to the present invention, and the scene data can be classified and extracted according to actual needs.

[0251] Those skilled in the art know that, in addition to implementing the system, device and each module thereof provided by the present invention in a pure computer readable program code manner, the same program can also be implemented in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers and embedded microcontrollers by logically programming the method steps. Therefore, the system, device and each module thereof provided by the present invention can be considered as a hardware component, and the modules included therein for implementing various programs can also be considered as structures in the hardware component; the modules for implementing various functions can also be considered as both software programs for implementing methods and structures in the hardware component.

[0252] The specific embodiments of the present invention are described above. It needs to be understood that the present invention is not limited to the specific embodiments described above, and various changes or modifications can be made by those skilled in the art within the scope of the claims, which does not affect the essential content of the present invention. The embodiments of the present application and the features in the embodiments can be arbitrarily combined with each other without conflict.

Claims

1. A vehicle data test method based on comfort evaluation, characterized by, Comprise: Step S1: Establish the subjective evaluation index and the objective evaluation index of the vehicle, respectively obtain the subjective score and the objective score of the comfort evaluation, and obtain the final comfort score after weighting and normalization; The objective evaluation index includes acceleration index, car sickness value and objective comfort value; The acceleration index includes: Effective weighted root mean square acceleration: where a w represents the effective weighted root mean square acceleration, T is the measurement period, a w (t) is the real-time equivalent acceleration, where is the original lateral, longitudinal and vertical acceleration collected by the vehicle, h is the frequency of change of acceleration, is a function of the variable h; Equivalent acceleration: where w i is a weighting factor, respectively represent the lateral, longitudinal and vertical effective weighted root mean square acceleration; Car sickness value: Wherein, T is the whole time of vibration occurrence; Objective comfort value a eq : a eq = f(a, MSDV z ) The subjective evaluation index includes the subjective evaluation comfort score of the feeling group of different ages and different genders; The objective comfort score a mentioned in the evaluation index eq And the subjective evaluation comfort score is normalized according to different weighting coefficients to obtain the final comfort score of the vehicle. Step S2: Build a comfort evaluation model through the collected vehicle motion data and the final comfort score; Specifically including: Build a vehicle comfort evaluation model through a neural network; According to the comfort score and the motion data of the vehicle, a vehicle comfort evaluation model is built through a neural network using a preset proportion of original data; Use the remaining original data to verify the comfort evaluation model; The verified comfort evaluation model is used for comfort evaluation of the vehicle; Step S3: According to the comfort evaluation model, the non-comfort scene data is classified and extracted from the vehicle motion data; Specifically including: Classify and extract the scene data below the comfort score threshold in the automatic driving process, which is used for algorithm analysis and iteration of the automatic driving system; The comfort evaluation model runs on the vehicle in real time to extract non-comfort scene data, which is analyzed by reading the recorded data packet form to obtain non-comfort scene data; According to the reasons for exceeding the comfort evaluation threshold, different data labels are given to the scene data; The classified scene data is used as the data basis for automatic driving algorithm analysis, algorithm logic and parameter adjustment; The adjusted automatic driving algorithm is verified by using the extracted scene data to realize the test closed loop of the classified scene data; Step S4: The classified scene data is used for automatic driving algorithm improvement and new automatic driving algorithm feasibility verification.

2. The vehicle data test method based on comfort evaluation according to claim 1, characterized by, In the step S1: Based on the ISO 2631 international standard, the comfort score of the vehicle is established; Build a comfort evaluation test scene, including an automatic driving road test scene and a scene during daily actual vehicle driving; Data collection: including objective data and subjective data: Objective data includes vehicle driving parameters in ISO 2631 standard, including vehicle acceleration in lateral, longitudinal and vertical directions, and vibration frequency corresponding to three accelerations, vehicle speed, steering wheel angle, vehicle angular velocity and vehicle operating environment temperature; The subjective data includes the comfort evaluation score given by people of different ages and different genders after riding an automatic driving vehicle; The evaluation index includes the objective evaluation index and the subjective evaluation index.

3. A vehicle data test system based on comfort evaluation, characterized by, The vehicle data test method based on comfort evaluation according to any one of claims 1-2, comprising: Data acquisition module: collect the original data of each module of the vehicle; Comfort evaluation module: build a comfort evaluation model, and apply the evaluation model to the comfort evaluation process of the vehicle in real time; Scene data extraction module: classify and extract non-comfort scene data from the vehicle motion data according to the comfort evaluation model; The scene data classification module: extracts the non-comfortable scene data classification, and respectively labels the scene data according to the reasons of the non-comfortable score; The algorithm analysis module: according to the classified scene data processed by the scene classification module, the logic verification and parameter adjustment of the driving algorithm and the feasibility verification of the new algorithm are carried out, and the closed loop of the test data is realized; The scene data storage module: the classified scene data is stored, which is convenient for algorithm analysis and feasibility verification of new algorithm.

4. A computer-readable storage medium storing a computer program, characterized in that, The computer program is executed by the processor to realize the steps of the vehicle data test method based on comfort evaluation in any one of claims 1 to 2.

5. A vehicle device based on comfort evaluation, characterized by, Comprise: A controller; The controller comprises the computer readable storage medium storing the computer program of claim 4, and the computer program is executed by the processor to realize the steps of the vehicle data test method based on comfort evaluation in any one of claims 1 to 2; or the controller comprises the vehicle data test system based on comfort evaluation in claim 3.

Images