Data generation system, data generation method and data generation program
By integrating market and test data through feature extraction, selection, and complementary data calculation, the system addresses the challenge of creating accurate and comprehensive test data for vehicle testing, enhancing data quality and reducing costs.
Patent Information
- Application Number
- JP2023501710
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2021-02-24
- Publication Date
- 2024-11-13
- Estimated Expiration
- 2041-02-24
AI Technical Summary
The challenge is to create highly accurate test data for vehicle testing while reducing development costs and increasing the comprehensiveness of test patterns, as mass-produced vehicles have lower sensor accuracy and fewer sensors, limiting the use of their data for evaluation and testing.
A data generation system that integrates market data from mass-produced vehicles with test data used during development, utilizing feature extraction, data selection, complementary data calculation, and integrated data generation to enhance accuracy and comprehensiveness.
This approach enables the creation of highly accurate test data that covers a broader range of test patterns, reducing development costs and improving the quality of test data for vehicle evaluation and testing.
Smart Images

Figure 0007582438000001 
Figure 0007582438000002 
Figure 0007582438000003
Abstract
Description
[Technical field]
[0001] The present invention relates to a data generation system, a data generation method, and a data generation program for linking a plurality of data to generate new data. [Background technology]
[0002] With the spread of connected cars, it is now possible to collect various data from mass-produced cars driving around town. This collected data is then used to provide services such as congestion prediction, car insurance, and fault diagnosis. The collected data can also be used for vehicle design, development, and testing.
[0003] For example, Patent Document 1 describes a predictive diagnostic device that diagnoses the signs of abnormalities in equipment. The device described in Patent Document 1 acquires speed, external environment, acceleration, GPS data, etc., using sensors installed inside the vehicle. Then, driving data immediately after the shipment of the vehicle is used as so-called teacher data for the vehicle in its normal state.
[0004] Non-Patent Document 1 describes a model-free analysis technique that can accurately determine the state of a system by comparing features mechanically extracted from time-series data. Patent Document 2 describes an invariant analysis that automatically extracts the associations between sensors based on machine learning, based on time-series data from multiple sensors. [Prior art documents] [Patent documents]
[0005] [Patent Document 1] JP 2016-146169 A [Patent Document 2] International Publication No. 2019 / 026193 [Non-patent literature]
[0006] [Non-Patent Document 1] Naoyuki Yoshinaga, Ryosuke Togawa, Ikuhiro Ajiro, "Model-free Analysis Technology for Time-series Data", NEC Technical Report, Vol. 72, No. 1, October 2019 Summary of the Invention [Problem to be solved by the invention]
[0007] On the other hand, as vehicle functions become more advanced and sophisticated, the number of patterns to be tested also increases, and the number of malfunctions that occur accordingly tends to increase. Therefore, it is preferable to be able to prepare data that covers all test patterns.
[0008] Here, various data in various driving environments can be obtained from mass-produced cars. However, in order to reduce costs, the number of sensors mounted on mass-produced cars is generally smaller than those used in testing, and the sensors are generally less accurate. Therefore, the data obtained directly from mass-produced cars is less accurate, and may not be fully usable as data for evaluation or testing. Therefore, sufficient evaluation or testing cannot be performed by simply using so-called data after the shipment of the automobile, as in the method described in Patent Document 1.
[0009] On the other hand, the data obtained during testing during development can be said to be highly accurate because of the high accuracy and variety of sensor values. However, when considering development costs, there are limitations to the test period and the patterns that can be considered, and the comprehensiveness of test patterns is often not necessarily sufficient.
[0010] Therefore, it is preferable to be able to generate highly accurate test data so as to increase the coverage of test patterns while suppressing development costs.
[0011] Therefore, an object of the present invention is to provide a data generation system, a data generation method, and a data generation program that can generate highly accurate test data so as to increase the coverage of test patterns while suppressing development costs. [Means for solving the problem]
[0012] The data generation system according to the present invention is a data generation system that generates new data using market data collected from mass-produced vehicles and test data used for testing vehicles in the development stage, and is characterized by comprising: a feature extraction means that extracts features of at least one of the market data and the test data; a data selection means that selects one or more other data including features corresponding to the features of the extracted one data; a complementary data calculation means that calculates complementary data that complements the market data or the test data from the one data and the selected other data; and an integrated data generation means that generates integrated data by integrating the calculated complementary data with at least one or both of the market data and the test data.
[0013] The data generation method according to the present invention is a data generation method for generating new data using market data collected from mass-produced vehicles and test data used for testing vehicles in the development stage, and is characterized in that a computer extracts features of at least one of the market data and the test data, selects one or more other data including features corresponding to the features of the extracted one data, calculates complementary data that complements the market data or the test data from the one data and the selected other data, and generates integrated data by integrating the calculated complementary data with at least one or both of the market data and the test data.
[0014] The data generation program according to the present invention is a data generation program applied to a computer that generates new data using market data collected from mass-produced vehicles and test data used for testing vehicles in the development stage, and is characterized in that it causes the computer to realize a feature extraction process that extracts features of at least one of the market data and the test data, a data selection process that selects one or more other data containing features corresponding to the features of the extracted one data, a complementary data calculation process that calculates complementary data that complements the market data or the test data from the one data and the selected other data, and an integrated data generation process that generates integrated data by integrating the calculated complementary data with at least one or both of the market data and the test data. Effect of the Invention
[0015] According to the present invention, highly accurate test data can be created so as to increase the coverage of test patterns while suppressing development costs. [Brief description of the drawings]
[0016] [Figure 1] 1 is a block diagram showing an example of the configuration of an embodiment of a data generation system according to the present invention; [Diagram 2] 11 is a flowchart illustrating an example of the operation of the data generating system. [Diagram 3] FIG. 11 is an explanatory diagram showing an example of learning data. [Figure 4] FIG. 13 is an explanatory diagram showing an example of market data. [Diagram 5] 1 is a block diagram showing an overview of a data generation system according to the present invention; [Figure 6] FIG. 1 is a schematic block diagram illustrating a configuration of a computer according to at least one embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0017] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
[0018] 1 is a block diagram showing an example of the configuration of an embodiment of a data generation system according to the present invention. The data generation system 100 of this embodiment includes a storage unit 10, a market data acquisition unit 20, a feature extraction unit 30, a data selection unit 40, a supplementary data calculation unit 50, and an integrated data generation unit 60.
[0019] The storage unit 10 stores various information used for processing by the data generation system 100 of this embodiment. Specifically, the storage unit 10 stores data collected from mass-produced vehicles (hereinafter referred to as market data) and data used for testing vehicles in the development stage (hereinafter referred to as test data). Mass-produced vehicles are vehicles that have completed the development stage and are mass-produced for sale in the market, and are vehicles that are actually operated and driven by consumers, etc.
[0020] The nature of market data and test data differs greatly depending on the environment in which they are collected. First, in terms of data volume, market data is obtained from mass-produced vehicles, so it is possible to collect a large amount of normal data. Generally, there is less defect data obtained from mass-produced vehicles compared to normal data. On the other hand, test data is smaller in volume than market data from the perspective of development costs, etc. (for example, confirmed tests are not performed multiple times, etc.).
[0021] Next, in terms of data accuracy, the accuracy of test data is generally high because vehicles usually have many types of sensors and are in an environment where data can be collected reliably. Also, test data can be created for each type of test, such as unit testing, integration testing, and driving testing. On the other hand, the accuracy of market data is generally lower than that of test data, because mass-produced vehicles usually have fewer types of sensors than those used during development and data loss is expected depending on communication conditions.
[0022] Another feature of market data is that various data is collected from multiple mass-produced vehicles. More specifically, market data includes data that is collected continuously, such as telematics data transmitted from connected cars, and data that is collected at specific times, such as DTC (Diagnostic Trouble Code) data extracted from ECU (Engine Control Unit) when a breakdown occurs.
[0023] Specifically, the driving data includes time-series data of sensor values acquired from various vehicle components such as an OBC (On-Board Charger) / CAN (Controller Area Network)-Bus, GPS (Global Positioning System) data, telematics data, etc. Also, the video data includes video captured by a drive recorder (e.g., a video of the front).
[0024] In addition, it is possible to obtain environmental information (e.g., weather information (such as typhoons and snow)) from the driving data that is difficult to collect in a test environment.
[0025] In addition to data acquired as driving data, malfunction data may also include malfunction reports (such as malfunction location, details, cause, and countermeasures) when the vehicle is brought to a dealer.
[0026] Depending on the specifications of a mass-produced vehicle, the acquired video and the failure data are not necessarily linked, but by associating the video with the driving data, the link may be established. One method for associating the video with the driving data is to tag the video.
[0027] On the other hand, the characteristics of test data include the high number and accuracy of sensors used for testing, and the ease of acquiring driving conditions at the time of data acquisition. However, due to limited time and resources from the perspective of development costs, it is difficult to cover all conditions with test data. Furthermore, even if no problems occur with the test data for unit testing, problems may occur in integration testing and driving testing from the perspective of comprehensiveness.
[0028] Specifically, the test data includes data similar to the driving data, and includes more items with higher accuracy than the market data. Also, the video data can include not only the video of the front view captured by the drive recorder, but also the video captured by the multi-directional camera and the video of the inside of the car.
[0029] Furthermore, test data is often created based on a comprehensive test specification in terms of test scenarios. Items included in the test specification (test data) include version (model number), individual number, inspection target (single unit (part), combination (assembly), integration (vehicle) and inspection viewpoint (functional / non-functional), as well as prerequisites (condition of other parts, driving environment, etc.), test procedure (control input, load input), expected result (normal / failure), judgment criteria (threshold, etc.), judgment result (OK / NG), and other items (reason for judgment, reason for exception, etc.).
[0030] The storage unit 10 stores market data acquired by a market data acquisition unit 20, which will be described later. The storage unit 10 may store market data acquired and created by other methods. The storage unit 10 also stores test data created by a designer or the like.
[0031] The market data acquiring unit 20 acquires market data collected from mass-produced vehicles and stores it in the storage unit 10. The market data acquiring unit 20 may acquire, for example, driving data or video data transmitted from a connected car equipped with a communication function. In addition, the market data acquiring unit 20 may improve the quality of the acquired market data by performing data cleansing such as converting the data into a code or deleting outliers.
[0032] The feature extraction unit 30 extracts features of at least one of the market data and the test data. That is, the feature extraction unit 30 may extract features of the market data or may extract features of the test data.
[0033] Here, the reason why the feature extraction unit 30 extracts features from the market data and the test data will be explained. The present invention aims to create more accurate data by taking advantage of the respective advantages of the market data and the test data and supplementing the information lacking in one data with the other data.
[0034] On the other hand, it is difficult to simply integrate market data and test data because the items and the environments in which the market data and test data were obtained are different. Therefore, in this embodiment, attention is paid to the characteristics of the data itself, and data with the same or similar characteristics are associated with each other to compensate for the missing data and create highly accurate data.
[0035] In the following description, in order to facilitate understanding of the invention, a case will be described in which the feature extraction unit 30 extracts features of market data from the market data. Note that the following process can also be applied to the case in which features are extracted from test data.
[0036] Various methods can be used as a method for the feature extraction unit 30 to extract features from market data. For example, the feature extraction unit 30 may extract data items themselves that represent the features of the data from the market data. Examples of such data items include the vehicle's individual number and vehicle model. Additionally, the feature extraction unit 30 may calculate the correlation between data items that indicate numerical data such as speed and acceleration, and extract the correlation between the data items as a feature.
[0037] Furthermore, the feature extraction unit 30 may calculate a feature amount that combines the relationship between the values of the sensors mounted on the vehicle and the association with the time-series changes in the values of the sensors, and extract the feature amount as a feature of a plurality of data. The feature extraction unit 30 may calculate such a feature amount by using, for example, a model-free analysis technique described in Non-Patent Document 1.
[0038] The method by which the feature extraction unit 30 extracts features from data is not limited to the above method. For example, when the market data is log data, the feature extraction unit 30 may extract the log pattern itself as a feature. Also, when the market data is a group of sensor data, the feature extraction unit 30 may extract the relationship between past sensor data as a feature using an invariant analysis technique as described in Patent Document 2.
[0039] The data selection unit 40 selects one or more pieces of data including features corresponding to the extracted features of the other data. For example, if features are extracted from market data, the data selection unit 40 selects one or more pieces of test data that match or are similar to the extracted features. On the other hand, if features are extracted from test data, the data selection unit 40 selects one or more pieces of market data that match or are similar to the extracted features.
[0040] The method by which the data selection unit 40 selects data is not particularly limited, and the content of the method is arbitrary as long as the method can select data with matching or similar characteristics. Specifically, the data selection unit 40 may predetermine items to be compared between the market data and the test data, and select data whose contents match or are within a predetermined range. For example, when the above-mentioned individual number, vehicle model, and correlation value of numerical data are determined as items to be compared, the data selection unit 40 may select data whose items match or are similar.
[0041] Furthermore, there may be a difference in the content of market data acquired from aged mass-produced vehicles and data acquired from new vehicles depending on the degree of deterioration. Therefore, the data selection unit 40 may apply a predetermined weighting to the characteristics of one of the data before comparing the characteristics of the data.
[0042] More specifically, the data selection unit 40 may calculate a weight value according to the degree of deterioration (for example, a weight value that changes the extracted feature more as the driving distance or driving time increases), multiply the feature by the calculated weight, and select the corresponding data by comparing the features. Note that any method for calculating the weight value may be used, and the method may be determined in advance according to the nature of the item. For example, the data selection unit 40 may determine that a weight of 0.8 is set for a feature of data acquired from a vehicle that has been driven for 10 years.
[0043] Furthermore, for example, when the above-described log pattern is extracted as a feature, the data selection unit 40 may compare the pattern with the market data log or the test data log to select corresponding data. Furthermore, for example, when the above-described relationship between past sensor data is extracted as a feature, the data selection unit 40 may compare the market data log or the test data log with the feature to select corresponding data.
[0044] Alternatively, a correspondence table for determining whether items to be compared match or are similar may be predefined, and the data selection unit 40 may select data whose items to be compared are defined in the correspondence table. In addition, when a relationship between time-series changes is extracted as a feature, the data selection unit 40 may select multiple data corresponding to that feature.
[0045] Furthermore, the data selection unit 40 may narrow down the selected test data to test data of a situation similar to the acquired market data. Examples of test data of a similar situation include test data of a similar sensor value and test data of a similar forward-view image. By narrowing down the test data to such test data, for example, when there is data (e.g., rearward-view image) that is not measured in the acquired market data but is included in the selected test data, it becomes possible to use the test data that is closer to the market data as supplementary data.
[0046] The complementary data calculation unit 50 calculates data (hereinafter referred to as complementary data) that complements the market data or the test data from the data whose features have been compared (one data) and the selected data (the other data).
[0047] Here, complementary data refers not only to data that complements missing data in either or both of the market data and test data, but also to data that refines existing data and new data that is generated to shorten the time interval between each data item.
[0048] In the following, a method will be described in which the complementary data calculation unit 50 calculates complementary data that complements market data from a plurality of selected test data. However, the following process can also be applied in the same way when test data is complemented from selected market data.
[0049] There are various ways to generate the complementary data. The first way is to complement missing market data items with test data. A specific example will be described below.
[0050] As a first specific example, there is a method of complementing missing items in the market data by using test data of similar driving scenes. In this case, the data selection unit 40 selects test data similar to the characteristics indicating the driving scenes of the market data. Then, the complementary data calculation unit 50 may identify the missing items in the market data, and extract items that are closest to the missing items in the market data from the items of the selected test data to generate complementary data.
[0051] The complementary data calculation unit 50 may calculate complementary data by using values before and after the test data collected in a time series (for example, by calculating an average value). In addition, the complementary data calculation unit 50 may calculate complementary data by using a missing data complementation method such as multiple imputation. Furthermore, the complementary data calculation unit 50 may calculate complementary data by using data of similar points in terms of time, speed, data tendency, etc. under the same conditions. As described above, the complementary data calculation unit 50 may generate integrated data after correcting data according to the degree of deterioration such as mileage. These methods can also be used in the following exemplary methods.
[0052] In this manner, the complementary data calculation unit 50 may extract data for an item missing from one data (e.g., market data) from the other selected data (e.g., test data) and calculate complementary data that complements the market data or the test data.
[0053] A second specific example is a method of complementing missing items in market data based on other correlations (e.g., correlations of other sensors). In this case, the data selection unit 40 selects test data similar to features extracted using a technique such as the model-free analysis described above. Then, the complementary data calculation unit 50 may generate complementary data that complements missing items in market data from the correlations of the selected test data.
[0054] For example, when features are extracted by the above-mentioned model-free analysis, the data selection unit 40 selects data including similar features from past test data. Then, the complementary data calculation unit 50 performs processing such as extraction of data corresponding to the interval to be complemented, normalization of the data, and biasing of the detected data, and calculates complementary data using the processed data.
[0055] Also, for example, when features are extracted by the above-described invariant analysis, the data selection unit 40 selects data related to the data to be complemented. Then, the complement data calculation unit 50 calculates the complement data by predicting the data to be complemented from the selected data using the relationship.
[0056] In the case of a sensor failure or the like, the complementary data calculation unit 50 may generate complementary data by combining findings from a knowledge base.
[0057] The second aspect is to improve the accuracy of market data by using test data, for example, a method of generating integrated data at 0.1 second intervals using market data collected at 1 second intervals. In this case, the data selection unit 40 may select test data similar to the characteristics showing the driving scene of the market data, and the supplementary data calculation unit 50 may generate new data by using the selected test data so as to shorten the time intervals of each data.
[0058] In this manner, the complementary data calculation section 50 may calculate complementary data for a time interval shorter than the time interval at which one data (e.g., market data) was collected, from the other selected data (e.g., test data).
[0059] Furthermore, the complementary data calculation unit 50 may change the method of calculating complementary data according to the characteristics of the data to be used. For example, the characteristics of the data to be used are classified into normal data and abnormal data. Since the abnormal data is considered to be data that should be given more importance than the normal data, the complementary data calculation unit 50 may calculate complementary data for the abnormal data in more detail than complementary data for the normal data.
[0060] Methods for performing detailed calculations include using the type of data to be calculated that represents more detailed information (for example, normal data is of type int and abnormal data is of type double), or shortening the time interval between created data (for example, normal data is of type every 1 second and abnormal data is of type 0.1 seconds), etc.
[0061] The integrated data generating unit 60 generates data (hereinafter referred to as integrated data) by integrating the calculated complementary data with at least one or both of the market data and the test data.
[0062] For example, when a missing item is calculated as complementary data, the integrated data generating unit 60 may generate integrated data in which the missing portion is filled by integrating the complementary data into the market data. Also, when new data is generated as complementary data to shorten the time interval between each piece of data, the integrated data generating unit 60 may insert the generated data into the existing market data to generate integrated data with a shorter time interval.
[0063] The market data acquisition unit 20, the feature extraction unit 30, the data selection unit 40, the complementary data calculation unit 50, and the integrated data generation unit 60 are realized by a computer processor (e.g., a CPU (Central Processing Unit)) that operates according to a program (a data generation program).
[0064] For example, a program may be stored in the storage unit 10, and the processor may read the program and operate as the market data acquisition unit 20, the feature extraction unit 30, the data selection unit 40, the supplementary data calculation unit 50, and the integrated data generation unit 60 in accordance with the program. Furthermore, the functions of the data generation system 100 may be provided in a SaaS (Software as a Service) format.
[0065] The market data acquisition unit 20, the feature extraction unit 30, the data selection unit 40, the supplementary data calculation unit 50, and the integrated data generation unit 60 may each be realized by dedicated hardware. Also, some or all of the components of each device may be realized by general-purpose or dedicated circuits, processors, etc., or a combination of these. These may be configured by a single chip, or may be configured by multiple chips connected via a bus. Some or all of the components of each device may be realized by a combination of the above-mentioned circuits, etc., and programs.
[0066] Furthermore, when some or all of the components of the data generation system 100 are realized by a plurality of information processing devices, circuits, etc., the plurality of information processing devices, circuits, etc. may be centrally arranged or distributed. For example, the information processing devices, circuits, etc. may be realized as a client-server system, a cloud computing system, etc., in a form in which each is connected via a communication network.
[0067] Next, the operation of the data generation system 100 of this embodiment will be described. Fig. 2 is a flowchart showing an example of the operation of the data generation system 100 of this embodiment. Here, it is assumed that market data acquired by the market data acquisition unit 20 and test data created by a designer or the like are stored in the storage unit 10.
[0068] The feature extraction unit 30 extracts features from at least one of the market data and the test data (step S11). The data selection unit 40 selects one or more other data including features corresponding to the features of the one data (step S12). The complementary data calculation unit 50 calculates complementary data that complements the market data or the test data from the one data and the selected other data (step S13). Then, the integrated data generation unit 60 generates integrated data by integrating the calculated complementary data with at least one or both of the market data and the test data (step S14).
[0069] As described above, in this embodiment, the feature extraction unit 30 extracts features from at least one of the market data and the test data, and the data selection unit 40 selects one or more other data including features corresponding to the features of the one data. The complementary data calculation unit 50 calculates complementary data that complements the market data or the test data from the one data and the selected other data, and the integrated data generation unit 60 generates integrated data by integrating the calculated complementary data with at least one or both of the market data and the test data. Thus, highly accurate test data can be created to increase the coverage of test patterns while suppressing development costs.
[0070] The specific operation of the data generation system of this embodiment will be described below. First, as a premise, the contents of the market data are confirmed by a designer or the like, and missing parts (for example, parts that are set to None, parts that are insufficient as time-series data, etc.) are identified.
[0071] Next, an analysis device (not shown) analyzes the features of the market data. The analysis device analyzes, for example, the slope and average value of the data (for example, the average value of the slope of X and Y coordinates), and indexes these features. When the above-mentioned model-free analysis is used, the analysis device trains a feature extraction engine using accumulated learning data. Then, the analysis device uses the trained feature extraction engine to generate binary feature data from the learning data. The generated feature data is stored in the storage unit 10.
[0072] FIG. 3 is an explanatory diagram showing an example of learning data. Data d1 and data d2 shown in FIG. 3 are part of test data collected in a time series during a driving test. For example, when the model-free analysis technique is used, binary data of
[0100] is generated as feature data from data d1, and binary data of
[1001] is generated as feature data from data d2, and these are stored in the storage unit 10. Note that this binary data is just an example.
[0073] Next, the market data acquisition unit 20 receives market data from the mass-produced vehicles and stores it in the storage unit 10. FIG. 4 is an explanatory diagram showing an example of market data. The market data d3 shown in FIG. 4 is missing some data for some reason, and the data d32 is "None". Moreover, the market data shown in FIG. 4 does not include the X-axis speed and the Y-axis speed, unlike the market data shown in FIG. 3.
[0074] The feature extraction unit 30 extracts features from the market data. The feature extraction unit 30 may calculate the data slope and average value as described above from the market data and extract them as features. The feature extraction unit 30 may also use the feature extraction engine to extract binary feature data from the market data illustrated in FIG. 4. For example, when the model-free analysis technique is used, the feature of the data d31 portion is converted to
[0100] , and the feature of the data d33 portion is converted to
[1000] .
[0075] The data selection unit 40 selects test data to be used for calculating the complementary data. Specifically, the data selection unit 40 matches the extracted market data features with the test data features and selects the test data with the highest similarity. For example, when the average of the X-coordinate slope and the average of the Y-coordinate slope of each test data are calculated as features, the data selection unit 40 may select the test data with the closest slope.
[0076] In addition, when the above-mentioned binary type feature data is generated, the data selection unit 40 may select the data d31 as the test data to be used for calculating the complementary data, since data matching the binary data
[0100] exists.
[0077] The complementary data calculation unit 50 calculates complementary data. For example, the complementary data calculation unit 50 may select two points from the data d31 and use the data of the two selected points as complementary data as is, or may calculate the average of the two points and use it as complementary data. The complementary data calculation unit 50 may also extract items (X-axis speed and Y-axis speed) that do not exist in the market data from the test data and use them as complementary data. In this way, the complementary data calculation unit 50 may complement the market data by using data before and after the missing data of the market eta or data of a similar portion of the test data.
[0078] Then, the integrated data generating unit 60 generates integrated data by integrating the calculated complementary data.
[0079] Next, application examples of the data generation system of this embodiment will be described. As a first application example, there is an application example in which multiple test data that match the characteristics of target market data are selected and complementary items are calculated. Specifically, when the market data acquisition unit 20 acquires market data collected from mass-produced vehicles, the data selection unit 40 selects multiple test data that match the characteristics extracted by the feature extraction unit 30.
[0080] The complementary data calculation unit 50 extracts data (e.g., slope, correlation, etc.) corresponding to the item to be complemented from the test data. The complementary data calculation unit 50 calculates a complementary value (e.g., average, median, mode, etc.) from the extracted data. Then, the integrated data generation unit 60 generates integrated data by integrating the calculated value with the market data.
[0081] As a second application example, multiple market data for an assumed situation are selected to increase the variety of data. Here, it is assumed that multiple market data acquired by the market data acquisition unit 20 are stored in the storage unit 10.
[0082] The feature extraction unit 30 extracts features used to identify a specified situation. The data selection unit 40 selects multiple market data that match the features extracted by the feature extraction unit 30. The supplementary data calculation unit 50 then calculates representative data from the selected multiple market data. Examples of methods for calculating the representative data include a method using statistical data such as the median, average, and mode of each item, and a method of randomly specifying the data.
[0083] Then, the complementary data calculation unit 50 calculates complementary values in the same manner as in the first application example in order to match the accuracy of the representative data with the accuracy of the test data, and the integrated data generation unit 60 generates integrated data by integrating the calculated values with the market data.
[0084] A third application example is an application example for making market data richer. For example, a sensor installed in a test vehicle may be removed from a mass-produced vehicle to reduce costs. For example, a mass-produced vehicle may be equipped with a sensor and a forward-facing camera for autonomous driving. On the other hand, a test vehicle may be equipped with not only a sensor and a forward-facing camera, but also a rear-facing camera for autonomous driving testing.
[0085] Therefore, the integrated data generating unit 60 integrates a portion of the market data of a similar situation with data not included in the market data. In the above example, the integrated data generating unit 60 integrates the image of the rear camera of the test data with the data of the market vehicle. This makes it easier to grasp the driving state of the mass-produced vehicle, and improves the analysis accuracy. For example, since it is possible to create test data for simulation showing a virtual surrounding situation, it is also possible to use this test data as learning image for image analysis AI (Artificial Intelligence).
[0086] The fourth application example is to reinforce the scenario of the test data. Specifically, a scenario that could not be executed with the test data is created by using market data. For example, it is possible to create a new test scenario by extracting market data related to an unexecuted test scenario and constructing test data.
[0087] Next, an overview of the present invention will be described. Fig. 5 is a block diagram showing an overview of a data generation system according to the present invention. The data generation system 80 according to the present invention is a data generation system (e.g., data generation system 100) that generates new data using market data collected from mass-produced vehicles and test data used for testing vehicles in the development stage, and includes a feature extraction means 81 (e.g., feature extraction unit 30) that extracts features of data from at least one of the market data and the test data, a data selection means 82 (e.g., data selection unit 40) that selects one or more other data (e.g., test data) including features corresponding to the features of the extracted one data (e.g., market data), a complementary data calculation means 83 (e.g., complementary data calculation unit 50) that calculates complementary data that complements the market data or the test data from the one data and the selected other data, and an integrated data generation means 84 (e.g., integrated data generation unit 60) that generates integrated data by integrating the calculated complementary data with at least one or both of the market data and the test data.
[0088] With such a configuration, highly accurate test data can be created while suppressing development costs and increasing the coverage of test patterns.
[0089] Specifically, the feature extraction means 81 may extract features of the market data from the market data, the data selection means 82 may select a plurality of test data corresponding to the extracted features of the market data, the complementary data calculation means 83 may calculate complementary data that complements the market data from the selected plurality of test data, and the integrated data generation means 84 may generate integrated data by integrating the calculated data with the market data.
[0090] Furthermore, the data selection means 82 may further select test data of a situation similar to the market data from the selected test data.
[0091] In addition, the feature extraction means 81 may calculate a feature that combines the relationship between the values of each sensor installed in the vehicle and the association with the time-series changes in the sensor values, and extract the calculated feature as a feature of multiple data.
[0092] Furthermore, the data selection means 82 may calculate a weight value according to the degree of deterioration, multiply the calculated weight by the feature, and compare the results to select the corresponding data.
[0093] The complementary data calculation means 83 may extract data for an item missing from one of the data from the other selected data, and calculate complementary data that complements the market data or the test data.
[0094] Furthermore, the complementary data calculation means 83 may calculate complementary data for a time interval shorter than the time interval at which one piece of data was collected, from the other selected piece of data.
[0095] 6 is a schematic block diagram showing the configuration of a computer according to at least one embodiment. A computer 1000 includes a processor 1001, a main memory device 1002, an auxiliary memory device 1003, and an interface 1004.
[0096] The above-mentioned data generation system 80 is implemented in a computer 1000. The operations of the above-mentioned processing units are stored in the form of a program (data generation program) in an auxiliary storage device 1003. The processor 1001 reads the program from the auxiliary storage device 1003, loads it in the main storage device 1002, and executes the above-mentioned processing in accordance with the program.
[0097] In at least one embodiment, the auxiliary storage device 1003 is an example of a non-transitory tangible medium. Other examples of non-transitory tangible media include a magnetic disk, a magneto-optical disk, a CD-ROM (Compact Disc Read-only memory), a DVD-ROM (Read-only memory), and a semiconductor memory connected via the interface 1004. In addition, when this program is distributed to the computer 1000 via a communication line, the computer 1000 that receives the program may load the program into the main storage device 1002 and execute the above-mentioned process.
[0098] The program may be for realizing part of the above-mentioned functions. Furthermore, the program may be a so-called differential file (differential program) that realizes the above-mentioned functions in combination with another program already stored in the auxiliary storage device 1003.
[0099] A part or all of the above-described embodiments can be described as, but is not limited to, the following supplementary notes.
[0100] (Appendix 1) A data generation system that generates new data using market data collected from mass-produced vehicles and test data used for testing vehicles in the development stage, A feature extraction means for extracting features of the data from at least one of the market data and the test data; a data selection means for selecting one or more other data including features corresponding to the extracted features of the one data; a complementary data calculation means for calculating complementary data that complements the market data or the test data from the one data and the selected other data; and an integrated data generating means for generating integrated data by integrating the calculated complementary data with at least one or both of the market data and the test data. A data generation system comprising:
[0101] (Appendix 2) The feature extraction means extracts features of the market data from the market data; The data selection means selects a plurality of test data corresponding to the extracted characteristics of the market data; The complementary data calculation means calculates complementary data that complements the market data from the selected plurality of test data; The integrated data generating means generates integrated data by integrating the calculated complementary data with the market data. 2. The data generation system of claim 1.
[0102] (Appendix 3) The data selection means further selects test data of a situation similar to the market data from the selected test data. 3. The data generation system of claim 2.
[0103] (Appendix 4) The feature extraction means calculates a feature quantity that combines the relationship between the values of each sensor mounted on the vehicle and the correlation with the time-series changes in the values of the sensors, and extracts the calculated feature quantity as a feature of multiple data. 4. A data generation system according to any one of claims 1 to 3.
[0104] (Appendix 5) The data selection means calculates a weight value according to the degree of deterioration, multiplies the calculated weight by the feature, and selects corresponding data by comparing the features. 5. A data generation system according to any one of claims 1 to 4.
[0105] (Appendix 6) The complementary data calculation means extracts data for items missing from one of the data from the other selected data, and calculates complementary data that complements the market data or the test data. 6. A data generation system according to any one of claims 1 to 5.
[0106] (Appendix 7) The complementary data calculation means calculates complementary data for a time interval shorter than the time interval at which one data was collected from the other selected data. 7. A data generation system according to any one of claims 1 to 6.
[0107] (Appendix 8) A data generation method for generating new data using market data collected from mass-produced vehicles and test data used for testing vehicles in the development stage, comprising: A computer extracts features of the data from at least one of the market data and the test data; The computer selects one or more other data items that include features corresponding to the extracted features of the one data item; The computer calculates complementary data that complements the market data or the test data from the one data and the selected other data, The computer generates integrated data by integrating the calculated complementary data with at least one or both of the market data and the test data. A data generation method comprising:
[0108] (Appendix 9) The computer extracts characteristics of the market data from the market data, The computer selects a plurality of test data corresponding to the extracted characteristics of the market data; The computer calculates complementary data that complements the market data from the selected multiple test data, The computer generates integrated data by integrating the calculated complementary data with the market data. 9. The method of generating data according to claim 8.
[0109] (Appendix 10) A program storage medium storing a data generation program applied to a computer that generates new data using market data collected from mass-produced vehicles and test data used for testing vehicles in the development stage, a feature extraction process for extracting features of the market data and / or the test data from the computer; a data selection process for selecting one or more other data items including features corresponding to the extracted features of the one data item; A complementary data calculation process for calculating complementary data that complements the market data or the test data from the one data and the selected other data; and An integrated data generation process for generating integrated data by integrating the calculated complementary data with at least one or both of the market data and the test data. A program storage medium storing a data generation program for realizing the above.
[0110] (Appendix 11) To the computer, A feature extraction process extracts features of the market data from the market data; In the data selection process, multiple test data corresponding to the extracted market data features are selected, In the complementary data calculation process, complementary data that complements the market data is calculated from the selected multiple test data. In an integrated data generation process, the calculated complementary data is integrated with the market data to generate integrated data. 11. The program storage medium of claim 10, storing a data generation program for the above.
[0111] (Appendix 12) A data generation program applied to a computer that generates new data using market data collected from mass-produced vehicles and test data used for testing vehicles in the development stage, a feature extraction process for extracting features of the market data and / or the test data from the computer; a data selection process for selecting one or more other data items including features corresponding to the extracted features of the one data item; A complementary data calculation process for calculating complementary data that complements the market data or the test data from the one data and the selected other data; and An integrated data generation process for generating integrated data by integrating the calculated complementary data with at least one or both of the market data and the test data. A data generation program to achieve this.
[0112] (Appendix 13) To the computer, A feature extraction process extracts features of the market data from the market data; In the data selection process, multiple test data corresponding to the extracted market data features are selected, In the complementary data calculation process, complementary data that complements the market data is calculated from the selected multiple test data. In an integrated data generation process, the calculated complementary data is integrated with the market data to generate integrated data. 13. The data generation program according to claim 12.
[0113] Although the present invention has been described above with reference to the embodiment, the present invention is not limited to the above embodiment. Various modifications that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention. [Industrial Applicability]
[0114] The present invention is preferably applied to a data generation system that generates new data by linking multiple data. Specifically, the present invention can be applied to solutions that use linked data. Examples of solutions that use linked data include failure sign detection, failure cause identification, deterioration prediction, and defect prediction. In addition, by collecting data under various environments and generating simulator data, it is also possible to contribute to the development of simulators. In addition, by generating data based on aging deterioration data and failure data of vehicles on the market, it is also possible to feed back this data to development. [Explanation of symbols]
[0115] 10 Storage section 20 Market Data Acquisition Department 30 Feature Extraction Unit 40 Data Selection Section 50 Complementary data calculation unit 60 Integrated Data Generation Division 100 Data Generation System
Claims
1. A data generation system that generates new data using market data collected from mass-produced vehicles and test data used for testing vehicles in a development stage, comprising: A feature extraction means for extracting features of the data from at least one of the market data and the test data; a data selection means for selecting one or more other data including features corresponding to the extracted features of the one data; a complementary data calculation means for calculating complementary data that complements the market data or the test data from the one data and the selected other data; and an integrated data generating means for generating integrated data by integrating the calculated complementary data with at least one or both of the market data and the test data. A data generation system comprising:
2. The feature extraction means extracts features of the market data from the market data; The data selection means selects a plurality of test data corresponding to the extracted characteristics of the market data; The complementary data calculation means calculates complementary data that complements the market data from the selected plurality of test data; The integrated data generating means generates integrated data by integrating the calculated complementary data with the market data. The data generation system of claim 1.
3. The data selection means further selects test data of a situation similar to the market data from the selected test data. The data generation system according to claim 2.
4. The feature extraction means calculates a feature amount that combines the relationship between the values of the sensors mounted on the vehicle and the association with the time-series changes in the values of the sensors, and extracts the calculated feature amount as a feature of the plurality of data. A data generation system according to any one of claims 1 to 3.
5. The data selection means calculates a weight value according to the degree of deterioration, multiplies the calculated weight by the feature, and selects corresponding data by comparing the features. A data generation system according to any one of claims 1 to 4.
6. The complementary data calculation means extracts data of an item missing from one of the data from the other selected data, and calculates complementary data that complements the market data or the test data. A data generation system according to any one of claims 1 to 5.
7. The complementary data calculation means calculates complementary data for a time interval shorter than the time interval at which one of the data was collected from the other of the selected data. A data generation system according to any one of claims 1 to 6.
8. A data generation method for generating new data using market data collected from mass-produced vehicles and test data used for testing vehicles in a development stage, comprising the steps of: A computer extracts features of the market data and / or the test data; The computer selects one or more other data items that include features corresponding to the extracted features of the one data item; The computer calculates complementary data that complements the market data or the test data from the one data and the selected other data, The computer generates integrated data by integrating the calculated complementary data with at least one or both of the market data and the test data. A data generation method comprising:
9. A computer extracts characteristics of the market data from the market data; The computer selects a plurality of test data corresponding to the extracted characteristics of the market data; The computer calculates complementary data that complements the market data from the selected multiple test data, The computer generates integrated data by integrating the calculated complementary data with the market data. The data generating method according to claim 8.
10. A data generation program applied to a computer that generates new data using market data collected from mass-produced vehicles and test data used for testing vehicles in a development stage, a feature extraction process for extracting features of the market data and / or the test data from the computer; a data selection process for selecting one or more other data items including features corresponding to the extracted features of the one data item; A complementary data calculation process for calculating complementary data that complements the market data or the test data from the one data and the selected other data; and An integrated data generation process for generating integrated data by integrating the calculated complementary data with at least one or both of the market data and the test data. A data generation program to achieve this.
Citation Information
Patent Citations
Lacking data complementing method and lacking data complementing system
JP2002215646A
Interactive 3D experiences based on data
JP2016146169A
System and method for unifying heterogenous datasets using primitives
US20200193324A1
Information processing device, information processing system, information processing method, and recording medium
WO2019026193A1
Cited By
Detection apparatus, detection method, and detection program
JP2024088196A
