Data collection method, data collection system, and model generation method
The data collection method and system address the challenge of diverse in-vehicle data collection by a center that acquires and creates datasets based on requirements, ensuring efficient data collection and model training readiness.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- DENSO CORP
- Filing Date
- 2025-11-26
- Publication Date
- 2026-06-04
AI Technical Summary
Existing systems face challenges in efficiently collecting diverse data from multiple in-vehicle devices due to the variety of data types handled by vehicles.
A data collection method and system that involves a center receiving data from in-vehicle devices, acquiring dataset requirements, and creating a dataset that satisfies these requirements by collecting and preprocessing vehicle data, with the ability to generate collection conditions and distribute them to devices for efficient data collection.
Enables efficient data collection from multiple in-vehicle devices by eliminating the need for users to identify necessary data, ensuring dataset completeness, and facilitating smooth transition to model training.
Smart Images
Figure JP2025041214_04062026_PF_FP_ABST
Abstract
Description
Data collection method, data collection system, and model generation method Cross-reference to related applications
[0001] This international application claims the benefit of Japanese Patent Application No. 2024-208265, filed with the Japan Patent Office on November 29, 2024, the entire disclosure of which is incorporated herein by reference.
[0002] This disclosure relates to a technique for collecting data from multiple vehicles.
[0003] Patent Document 1 describes a system for collecting vehicle data from a plurality of in-vehicle devices mounted on each of a plurality of vehicles.
[0004] Japanese Patent Application Laid-Open No. 2023-84379
[0005] As a result of the inventors' detailed examination, it has been found that there is a problem in that it is necessary to efficiently collect data from a plurality of in-vehicle devices because the data handled by vehicles is diverse.
[0006] This disclosure efficiently collects data from a plurality of in-vehicle devices.
[0007] One aspect of this disclosure is a data collection method executed at a center configured to receive vehicle data including at least information regarding the mounted vehicle from a plurality of in-vehicle devices mounted on each of a plurality of vehicles.
[0008] The center acquires dataset requirement data indicating dataset requirements for defining a dataset including a plurality of vehicle data used in model training to be performed on a machine learning model.
[0009] The center collects a plurality of vehicle data from a plurality of in-vehicle devices and creates a dataset that satisfies the dataset requirements indicated by the acquired dataset requirement data using the collected plurality of vehicle data.
[0010] In the data collection method of this disclosure configured in this way, the user registers the dataset requirements with the center, which then collects multiple vehicle data from multiple in-vehicle devices to create a dataset that satisfies the dataset requirements. Therefore, the data collection method of this disclosure eliminates the need for the user to identify the vehicle data necessary to complete the dataset by checking the contents of an incomplete dataset and then having the center acquire the identified necessary vehicle data, thereby enabling efficient data collection from multiple in-vehicle devices.
[0011] Another aspect of this disclosure is a data acquisition system comprising a plurality of in-vehicle devices and a center. The plurality of in-vehicle devices are installed in each of a plurality of vehicles and are configured to transmit vehicle data which includes at least information about the vehicle in which they are installed. The center is configured to receive vehicle data from the plurality of in-vehicle devices.
[0012] The center comprises an acquisition unit and a dataset creation unit.
[0013] The acquisition unit is configured to acquire dataset requirement data that indicates dataset requirements, which define a dataset containing multiple vehicle data sets used in model training for a machine learning model.
[0014] The dataset creation unit is configured to collect multiple vehicle data from multiple in-vehicle devices and use the collected vehicle data to create a dataset that satisfies the dataset requirements indicated by the acquired dataset requirements data.
[0015] The data collection system described herein is a system that implements the data collection method described herein and can achieve the same effects as the data collection method described herein.
[0016] Another aspect of this disclosure is a model generation method in which the center generates a machine learning model by performing model training using a dataset created by the data collection method of this disclosure.
[0017] The model generation method described herein is a method that uses the data collection method described herein, and by performing this method, the same effects as the data collection method described herein can be obtained.
[0018] This is a block diagram showing the configuration of the data collection system. This is a block diagram showing the configuration of the data collection device. This is a block diagram showing the configuration of the center. This is a functional block diagram showing the functional configuration of the center. This is a diagram illustrating the management of collection condition data and collected data. This is a diagram illustrating the registration of dataset requirements. This is a diagram illustrating the incorporation of data into the dataset and the generation of collection condition data. This is a diagram illustrating the incorporation of newly collected data into the dataset. This is a flowchart illustrating the dataset construction process.
[0019] Embodiments of the present disclosure will be described below with reference to the drawings.
[0020] As shown in Figure 1, the data acquisition system 1 of this embodiment comprises a plurality of data acquisition devices 2 and a center 3.
[0021] The data collection device 2 is mounted on the vehicle and has the function of communicating data with the center 3 via a wide-area wireless communication network NW.
[0022] Center 3 is a device that manages the data acquisition system 1. Center 3 has the function of communicating data with multiple data acquisition devices 2 via a wide-area wireless communication network NW.
[0023] As shown in Figure 2, the data acquisition device 2 comprises a control unit 11, a CAN communication unit 12, a storage unit 13, and a communication unit 14. CAN stands for Controller Area Network.
[0024] The control unit 11 is an electronic control device centered around a microcomputer equipped with a CPU 21, ROM 22, RAM 23, etc. The various functions of the microcomputer are realized by the CPU 21 executing a program stored in a non-transitional physical recording medium. In this example, the ROM 22 corresponds to the non-transitional physical recording medium that stores the program. Furthermore, the execution of this program executes a method corresponding to the program. Note that some or all of the functions executed by the CPU 21 may be configured hardware-wise by one or more ICs, etc. Also, the number of microcomputers constituting the control unit 11 may be one or more.
[0025] The CAN communication unit 12 is connected to multiple ECUs via communication lines to enable data communication and transmits and receives data according to the CAN communication protocol. Specifically, the multiple ECUs connected to the CAN communication unit 12 include an engine ECU for engine control, a brake ECU for brake control, a steering ECU for steering control, a suspension ECU for suspension control, and an ECU for controlling the on / off state of the lights. In Figure 2, only ECUs 16, 17, and 18 are shown as ECUs connected to the CAN communication unit 12. ECU stands for Electronic Control Unit.
[0026] The memory unit 13 is a memory device for storing various types of data.
[0027] The communications unit 14 performs data communication with the center 3 via the wide-area wireless communication network NW.
[0028] As shown in Figure 3, Center 3 comprises a control unit 31, a communication unit 32, and a storage unit 33.
[0029] The control unit 31 is an electronic control device centered around a microcomputer equipped with a CPU 41, ROM 42, RAM 43, etc. The various functions of the microcomputer are realized by the CPU 41 executing a program stored in a non-transitional physical recording medium. In this example, the ROM 42 corresponds to the non-transitional physical recording medium that stores the program. Furthermore, the execution of this program executes a method corresponding to the program. Note that some or all of the functions executed by the CPU 41 may be configured hardware-wise by one or more ICs, etc. Also, the number of microcomputers constituting the control unit 31 may be one or more.
[0030] The communication unit 32 performs data communication with multiple data acquisition devices 2 via a wide-area wireless communication network NW.
[0031] The storage unit 33 is a storage device for storing various types of data. The storage unit 33 is equipped with a collection condition database (hereinafter referred to as the collection condition DB) 33a and a collection data database (hereinafter referred to as the collection data DB) 33b.
[0032] Center 3, as a functional block realized by the CPU 41 executing a program stored in ROM 42, includes, as shown in Figure 4, a data collection unit 101, a collected data / product storage unit 102, a collected data management / analysis unit 103, a machine learning unit 104, a collection condition management unit 105, a CI / CD unit 106, a UI unit 107, authentication / authorization units 108, 109, 110, 111, 112, a collection condition file transmission unit 113, and an AI model file transmission unit 114. CI stands for Continuous Integration. CD stands for Continuous Delivery. UI stands for User Interface. AI stands for Artificial Intelligence.
[0033] The data acquisition unit 101 collects data from multiple data acquisition devices 2 installed in each of the multiple vehicles.
[0034] The collected data and product storage unit 102 stores the data collected by the data collection unit 101 and the products generated by the collected data management and analysis unit 103, the machine learning unit 104, and the collection condition management unit 105.
[0035] The collected data management and analysis unit 103 includes a preprocessing unit 121 and a data collection status confirmation unit 122.
[0036] The preprocessing unit 121 performs preprocessing on the collected data to make it easier to use in machine learning (for example, changing the resolution, changing to grayscale, etc.).
[0037] The data collection status confirmation unit 122 checks the amount of data collected.
[0038] The machine learning unit 104 includes an annotation unit 123, a model training unit 124, and a model evaluation unit 125.
[0039] The annotation unit 123 performs annotation on the data that has been preprocessed by the preprocessing unit 121, adding information (for example, labels) for machine learning.
[0040] The model training unit 124 trains a machine learning model using the data generated by the annotation unit 123. The machine learning model is, for example, a model that takes image data captured by an in-vehicle camera as input data, determines objects in the image, and outputs the determination result.
[0041] The model evaluation unit 125 evaluates the accuracy of the machine learning model based on the training results from the model training unit 124.
[0042] The data collection condition management unit 105 manages the data collection conditions used to collect data for training machine learning models. The data collection condition management unit 105 includes a data collection condition generation unit 126.
[0043] The data collection condition generation unit 126 generates data collection condition data that indicates the data collection conditions for collecting data to improve the accuracy of the machine learning model, based on the evaluation results from the model evaluation unit 125.
[0044] The CI / CD unit 106 includes a collection condition build and distribution unit 127, an AI logic build and distribution unit 128, and a distribution status management unit 129.
[0045] The collection condition build and distribution unit 127 converts the collection condition data generated by the collection condition generation unit 126 into a format recognizable by the data collection device 2 and distributes it to the data collection device 2.
[0046] The AI logic build and distribution unit 128 converts the machine learning model (i.e., AI logic) generated by the machine learning unit 104 into a format executable by the data collection device 2 and distributes it to the data collection device 2.
[0047] The distribution status management unit 129 checks the distribution status of the collection condition data and the AI logic.
[0048] The UI unit 107 is a device for information exchange between data engineers, AI engineers, software engineers, test engineers, operators and the center 3.
[0049] A data engineer is an engineer related to data used in machine learning. An AI engineer is an engineer related to the generation of machine learning models. A software engineer is an engineer related to applications generated by combining multiple machine learning models. A test engineer is an engineer who evaluates machine learning models and applications. An operator manages the distribution status of collection condition data and AI logic.
[0050] The authentication and authorization unit 108 performs authentication and authorization for the access of data engineers to the center 3. For example, the authentication and authorization unit 108 performs authentication and authorization so that data engineers can only access the collection data management and analysis unit 103.
[0051] The authentication and authorization unit 109 performs authentication and authorization for the access of AI engineers to the center 3. For example, the authentication and authorization unit 109 performs authentication and authorization so that AI engineers can only access the machine learning unit 104.
[0052] The authentication and authorization unit 110 authenticates and authorizes access to Center 3 by software engineers. For example, the authentication and authorization unit 110 authenticates and authorizes software engineers to access only the machine learning unit 104.
[0053] The authentication and authorization unit 111 authenticates and authorizes the test engineer's access to Center 3. For example, the authentication and authorization unit 111 authenticates and authorizes the test engineer to access only the data collection condition management unit 105.
[0054] The authentication and authorization unit 112 authenticates and authorizes the operator's access to the center 3. For example, the authentication and authorization unit 112 authenticates and authorizes the operator so that they can access only the CI / CD unit 106.
[0055] The collection condition file transmission unit 113 transmits the collection condition file generated by the collection condition build and distribution unit 127 to multiple data acquisition devices 2.
[0056] The AI model file transmission unit 114 transmits the AI model file generated by the AI logic build and distribution unit 128 to multiple data acquisition devices 2.
[0057] Furthermore, the authentication and authorization units 108-112 are configured to allow different access permissions to be set for each of the multiple applications (for example, ADAS applications, multimedia applications) that utilize the machine learning model handled by Center 3. ADAS stands for Advanced Driver Assistance System.
[0058] The data collection conditions are either manually generated by engineers or automatically generated by an AI model, as shown by arrows L1 and L2 in Figure 5.
[0059] The data collection conditions file shown in Figure 5 includes conditions C1, C2, and C3, etc. Conditions C1 and C2 are the conditions for executing data collection. Data collection is executed when both conditions C1 and C2 are met.
[0060] Condition C3 indicates the timing for uploading the collected data to Center 3.
[0061] A collection condition profile is attached to the collection condition file. The collection condition profile includes a collection condition ID and a scene tag. ID stands for identification. "CJD-001" shown in Figure 5 is the collection condition ID. "night rain" shown in Figure 5 is the scene tag. A scene includes ambient brightness, vehicle behavior, driver actions, road conditions, application execution status, etc. Therefore, the collection condition ID corresponds to a scene. A scene tag is a string set by engineers to identify a scene.
[0062] The collection condition file to which the collection condition profile has been added is stored in the collection condition DB33a, as indicated by arrow L3.
[0063] The data collection conditions file is distributed to the data collection device 2 mounted on the vehicle, as indicated by arrow L4.
[0064] The data acquisition device 2 acquires image data and sensor data, at least one of them, based on the acquisition condition file obtained from the center 3. Data D1 shown in Figure 5 is image data. Data D2 shown in Figure 5 is sensor data showing the detection results of sensors mounted on the vehicle.
[0065] The data collection device 2 adds the collection condition ID included in the collection condition profile corresponding to the collection condition file to the collected data. As shown by arrow L5, the data collection device 2 uploads the data with the added collection condition ID to the center 3.
[0066] Center 3 stores the data uploaded from data acquisition device 2 in the collected data DB 33b. As shown by arrow L6, the data stored in the collected data DB 33b is linked to scenes by the collection condition ID. In this way, the data stored in the collected data DB 33b is linked to the scene tag corresponding to the collection condition ID (for example, "night rain" in Figure 5).
[0067] The data stored in the collected data DB33b is used by engineers, as indicated by arrow L7.
[0068] As shown in Figure 6, the user (for example, a test engineer) creates dataset requirement data indicating the dataset requirements for each machine learning model by performing input operations via the UI unit 107, and registers the created dataset requirements. Center 3 registers the dataset requirements by saving the dataset requirement data to the storage unit 33.
[0069] Dataset requirements include the target scenes, the number of data points, the search criteria, and the deadline for creating the dataset.
[0070] The target scenes are those used for model training, such as "daytime," "evening," and "night."
[0071] The number of data points is the number of data points corresponding to the target scene, and is set for each target scene. In Figure 6, the number of data points for the target scenes "Daytime," "Evening," and "Night" is 3 each. Note that the number of data points may also be set based on the data distribution. For example, multiple numbers may be set for each hour from 0:00 to 24:00, such as A data points from 0:00 to 1, B data points from 1:00 to 2:00, ..., C data points from 13:00 to 14:00, ..., and D data points from 23:00 to 24:00.
[0072] Search criteria are conditions used to search for and extract data necessary for building a dataset from multiple collected data sets. Search criteria are set using, for example, vehicle type, data collection period, and required preprocessing. Vehicle type refers to the type of vehicle from which data is collected. Data collection period refers to the period during which data is collected. Required preprocessing refers to the preprocessing performed on the data before model training.
[0073] The dataset construction deadline is the deadline for constructing a dataset that corresponds to the dataset requirements data.
[0074] As shown in Figure 7, Center 3 had already acquired three data points for the target scene "daytime," two data points for the target scene "evening," and one data point for the target scene "night" as collected data at the time the dataset requirements were registered.
[0075] Center 3 searches through multiple collected datasets for data that matches the registered dataset requirements and incorporates the extracted data into the dataset.
[0076] In Figure 7, three data points are incorporated into the dataset for the target scene "daytime," two data points for the target scene "evening," and one data point for the target scene "night."
[0077] As a result, the target scene "Daytime" satisfies the dataset requirement in terms of the number of data points, while the target scenes "Evening" and "Night" do not.
[0078] Therefore, Center 3 generates collection condition data for collecting data for the target scenes "evening" and "night," and distributes the generated collection condition data to the vehicles that are to be distributed.
[0079] As shown in Figure 8, Center 3 acquires the data uploaded from the vehicle based on the new collection conditions as collected data.
[0080] Center 3 searches for data that matches the registered dataset requirements from the newly acquired collected data, and incorporates the extracted data into the dataset.
[0081] In Figure 8, one data point is added for the target scene "Evening," and two data points are added for the target scene "Night." As a result, the target scenes "Evening" and "Night" now meet the required number of data points for the dataset. Therefore, the dataset corresponding to the registered dataset requirements is completed.
[0082] Next, the procedure for the dataset construction process executed by Center 3 will be explained. The dataset construction process is a process that is repeatedly executed while the control unit 31 is operating.
[0083] When the dataset construction process is executed, the CPU 41 of the control unit 31 determines in S10, as shown in Figure 9, whether or not the dataset requirements have been registered via the UI unit 107. If the dataset requirements have not been registered, the CPU 41 terminates the dataset construction process.
[0084] On the other hand, if a dataset requirement is registered, in S20 the CPU 41 searches for data from among multiple collected data that matches the dataset requirement registered in S10 (hereinafter referred to as the target dataset requirement), and extracts the data that matches the target dataset requirement.
[0085] In S30, CPU 41 incorporates the data extracted in S20 into a dataset that corresponds to the target dataset requirements (hereinafter referred to as the target dataset).
[0086] In S40, the CPU 41 determines whether the target dataset satisfies the target dataset requirements. Specifically, the CPU 41 determines that the target dataset satisfies the target dataset requirements if the number of data points is sufficient for all target scenes included in the target dataset requirements.
[0087] If the target dataset satisfies the target dataset requirements, the CPU 41 registers the target dataset as a completed dataset (hereinafter referred to as a completed dataset) in S50.
[0088] In S60, CPU 41 notifies the user that the dataset construction is complete and terminates the dataset construction process.
[0089] Furthermore, in S40, if the target dataset does not satisfy the target dataset requirements, the CPU 41 generates collection condition data in S70 that indicates the collection conditions for collecting data for the target scene that does not meet the data requirements.
[0090] CPU 41, in S80, distributes the collection condition data generated in S70 to the vehicles to be distributed.
[0091] In S90, the CPU 41 determines whether or not new data has been uploaded from the vehicle. If no new data has been uploaded, the CPU 41 repeats the process in S90 and waits until new data is uploaded. Once new data is uploaded, in S100, the CPU 41 searches the newly uploaded data for data that matches the requirements of the target dataset, and if it is able to extract data that matches the requirements of the target dataset, it incorporates the extracted data into the target dataset.
[0092] In S110, the CPU 41 determines whether the target dataset satisfies the target dataset requirements. If the target dataset does not satisfy the target dataset requirements, the CPU 41 determines in S120 whether the deadline for constructing the dataset included in the target dataset requirements has expired.
[0093] If the deadline for dataset construction has not yet passed, CPU 41 proceeds to S90. On the other hand, if the deadline for dataset construction has passed, CPU 41 generates relaxed collection condition data in S130. Specifically, CPU 41 relaxes the collection conditions by increasing the number of vehicle types targeted for data collection or by extending the time period for data collection.
[0094] In S140, CPU 41 distributes the collection condition data generated in S130 to the vehicles to be distributed, and then proceeds to S90.
[0095] Furthermore, if the target dataset satisfies the target dataset requirements in S110, the CPU 41 registers the target dataset as a completed dataset in S150.
[0096] In S160, CPU 41 notifies the user that the dataset construction is complete, in the same manner as in S60, and terminates the dataset construction process.
[0097] The data acquisition system 1 configured in this way comprises a plurality of data acquisition devices 2 and a center 3.
[0098] Multiple data acquisition devices 2 are installed in each of the multiple vehicles and are configured to transmit vehicle data that includes at least information about the vehicle on which they are installed. Center 3 is configured to receive vehicle data from the multiple data acquisition devices 2.
[0099] Center 3 obtains dataset requirements data that defines the dataset requirements for a dataset containing multiple vehicle data sets used in model training for machine learning models.
[0100] Center 3 collects multiple vehicle data from multiple data acquisition devices 2, and uses the collected vehicle data to create a dataset that satisfies the dataset requirements indicated by the acquired dataset requirements data.
[0101] In this data collection system 1, the user registers the dataset requirements with the center 3, which then collects multiple vehicle data from multiple data collection devices 2 to create a dataset that satisfies the dataset requirements. Therefore, the data collection system 1 eliminates the need for the user to identify the vehicle data necessary to complete the dataset by checking the contents of an incomplete dataset and then having the center 3 acquire that identified vehicle data, thus enabling efficient data collection from multiple data collection devices 2.
[0102] Center 3 also stores vehicle data received from multiple data acquisition devices 2.
[0103] Center 3 creates a dataset by incorporating vehicle data that meets the dataset requirements from among multiple stored vehicle data sets.
[0104] If Center 3 is unable to complete a dataset that satisfies the dataset requirements using the multiple vehicle data stored, it will continue to collect multiple vehicle data from multiple data acquisition devices 2 until the dataset satisfies the dataset requirements.
[0105] Center 3 executes a migration process to move the dataset to model training once the dataset satisfies the dataset requirements. In this embodiment, the migration process is to notify the user that the dataset construction is complete.
[0106] Such a data collection system 1 can smoothly transition to model training using the completed dataset once the dataset is complete.
[0107] Furthermore, if Center 3 is unable to complete a dataset that satisfies the dataset requirements using multiple stored vehicle data, it generates collection condition data indicating the collection conditions for collecting the vehicle data necessary to satisfy the dataset requirements, and distributes the generated collection condition data to multiple data collection devices 2.
[0108] Since this data collection system 1 can collect vehicle data from data collection devices 2 only to the extent necessary to satisfy the dataset requirements, it can collect data more efficiently from multiple data collection devices 2.
[0109] Furthermore, when Center 3 obtains dataset requirement data, it determines whether the dataset can be completed using the stored vehicle data. If it cannot be completed using the stored vehicle data, it collects the necessary vehicle data to satisfy the dataset requirements and stores the vehicle data collected from the multiple data collection devices 2. This cycle is repeated sequentially until the dataset is completed. As a result, the data collection system 1 can minimize the user's input of dataset requirement data, thereby improving user convenience.
[0110] Furthermore, Center 3 accelerates data collection from multiple data collection devices 2 when a pre-set acceleration condition is met during the repetition of the cycle. In this embodiment, the acceleration condition is the expiration of the deadline for constructing the dataset. In this embodiment, Center 3 accelerates data collection by generating relaxed collection condition data. This prevents the data collection system 1 from remaining in a state where it cannot complete the dataset for an extended period of time.
[0111] Center 3 also obtains data indicating the acceleration conditions (i.e., data indicating the deadline for dataset construction) from the user via UI unit 107.
[0112] Center 3 also acquires dataset requirement data that defines the dataset containing multiple vehicle data to be used in model training for the machine learning model. Center 3 collects multiple vehicle data from multiple data collection devices 2 and uses the collected vehicle data to create a dataset that satisfies the dataset requirements indicated by the acquired dataset requirement data. Center 3 generates a machine learning model by performing model training using the dataset.
[0113] Since such a data collection system 1 can efficiently collect data from multiple data collection devices 2, it can efficiently generate machine learning models.
[0114] In the embodiment described above, the data acquisition device 2 corresponds to an in-vehicle device, S10 corresponds to processing as an acquisition unit, S20 to S160 correspond to processing as a data set creation unit, S60 and S160 correspond to migration processing, and the UI unit 107 corresponds to a user interface.
[0115] Although one embodiment of the present disclosure has been described above, the present disclosure is not limited to the above embodiment and can be implemented in various modified forms.
[0116] The control units 11, 31 and their methods described in this disclosure may be implemented by a dedicated computer provided by configuring a processor and memory programmed to execute one or more functions embodied by a computer program. Alternatively, the control units 11, 31 and their methods described in this disclosure may be implemented by a dedicated computer provided by configuring a processor by one or more dedicated hardware logic circuits. Alternatively, the control units 11, 31 and their methods described in this disclosure may be implemented by one or more dedicated computers configured by a combination of a processor and memory programmed to execute one or more functions and a processor configured by one or more hardware logic circuits. Furthermore, the computer program may be stored as instructions executed by the computer on a computer-readable non-transitional tangible recording medium. The methods for realizing the functions of each part included in the control units 11, 31 do not necessarily need to include software, and all of its functions may be realized using one or more hardware components.
[0117] Multiple functions of one component in the above embodiment may be realized by multiple components, or one function of one component may be realized by multiple components. Furthermore, multiple functions of multiple components may be realized by one component, or one function realized by multiple components may be realized by one component. Also, some parts of the configuration of the above embodiment may be omitted. Furthermore, at least some parts of the configuration of the above embodiment may be added to or replaced with the configuration of other above embodiments.
[0118] In addition to the data acquisition device 2 and center 3 described above, this disclosure can also be realized in various forms, such as a system comprising the data acquisition device 2 and center 3, a program for causing a computer to function as the data acquisition device 2 and center 3, a non-transitional physical recording medium such as a semiconductor memory on which this program is recorded, and a data acquisition method. [Technical Concept Disclosed in This Specification] [Item 1] A data acquisition method performed in a center (3) configured to receive vehicle data, which includes at least information about the vehicle, from a plurality of in-vehicle devices (2) mounted on each of a plurality of vehicles, the method comprising: acquiring dataset requirement data indicating dataset requirements that define a dataset including a plurality of the vehicle data used in model training performed on a machine learning model; collecting a plurality of the vehicle data from the plurality of in-vehicle devices; and using the collected plurality of vehicle data, creating a dataset that satisfies the dataset requirements indicated by the acquired dataset requirement data.
[0119] [Item 2] A data collection method according to Item 1, comprising: storing the vehicle data received from a plurality of the in-vehicle devices; creating the dataset by incorporating the vehicle data that matches the dataset requirements from the stored plurality of vehicle data into the dataset; if it is not possible to complete the dataset that satisfies the dataset requirements using the stored plurality of vehicle data, continuing to collect the plurality of vehicle data from the plurality of in-vehicle devices until the dataset satisfies the dataset requirements; and when the dataset satisfies the dataset requirements, executing a transition process (S60, S160) to proceed to model training.
[0120] [Item 3] A data collection method according to Item 2, wherein if a dataset that satisfies the dataset requirements cannot be completed using multiple stored vehicle data, the method generates collection condition data indicating collection conditions for collecting the vehicle data necessary to satisfy the dataset requirements, and distributes the generated collection condition data to multiple in-vehicle devices.
[0121] [Item 4] A data collection method according to Item 2 or Item 3, wherein, upon obtaining the dataset requirement data, the method determines whether the dataset can be completed using a plurality of stored vehicle data, and if it cannot be completed using the plurality of stored vehicle data, the method sequentially repeats the cycle of collecting the vehicle data necessary to satisfy the dataset requirement and storing the vehicle data collected from a plurality of in-vehicle devices until the dataset is completed.
[0122] [Item 5] A data collection method described in Item 4, wherein if a predetermined acceleration condition is met during the repetition of the cycle, the data collection from multiple in-vehicle devices is accelerated.
[0123] [Item 6] A data collection method described in Item 5, wherein data indicating the promotion conditions is obtained from a user via a user interface (107).
[0124] [Item 7] A data collection method according to any one of Items 1 to 6, wherein the dataset requirement includes at least one of a target scene which is the scene to be used for model training, the number of vehicle data corresponding to the target scene, the distribution of the vehicle data corresponding to the target scene, the type of vehicle from which the vehicle data is to be collected, the period for which the vehicle data is collected, and preprocessing performed on the vehicle data before the model training is performed.
[0125] [Item 8] A data collection system (1) comprising: a plurality of in-vehicle devices (2) mounted on each of a plurality of vehicles and configured to transmit vehicle data including at least information about the vehicle on which they are mounted; and a center (3) configured to receive the vehicle data from the plurality of in-vehicle devices, wherein the center comprises: an acquisition unit (S10) configured to acquire dataset requirement data indicating dataset requirements that define a dataset including a plurality of the vehicle data used in model training performed on a machine learning model; and a dataset creation unit (S20 to S160) configured to collect a plurality of the vehicle data from the plurality of in-vehicle devices and to create a dataset that satisfies the dataset requirements indicated by the acquired dataset requirement data using the plurality of the vehicle data.
[0126] [Item 9] A model generation method comprising: a center (3) configured to receive vehicle data from multiple in-vehicle devices (2) installed in each of multiple vehicles, each containing at least information about the vehicle on which the device is installed, and which obtains dataset requirement data indicating dataset requirements that define a dataset containing multiple vehicle data to be used in model training for a machine learning model; the center collecting multiple vehicle data from the multiple in-vehicle devices, and using the collected multiple vehicle data to create a dataset that satisfies the dataset requirements indicated by the obtained dataset requirement data; and the center generating the machine learning model by performing model training using the dataset.
Claims
1. A data collection method performed in a center (3) configured to receive vehicle data from multiple in-vehicle devices (2) installed in each of multiple vehicles, the center (3) being configured to receive vehicle data including at least information about the vehicle on which the device is installed, the method comprising: acquiring dataset requirement data indicating dataset requirements that define a dataset including multiple vehicle data to be used in model training for a machine learning model; collecting multiple vehicle data from the multiple in-vehicle devices; and using the collected multiple vehicle data to create a dataset that satisfies the dataset requirements indicated by the acquired dataset requirement data.
2. A data collection method according to claim 1, comprising: storing the vehicle data received from a plurality of in-vehicle devices; creating a dataset by incorporating vehicle data that matches the dataset requirements from among the stored plurality of vehicle data into the dataset; if it is not possible to complete the dataset that satisfies the dataset requirements using the stored plurality of vehicle data, continuing to collect the plurality of vehicle data from the plurality of in-vehicle devices until the dataset satisfies the dataset requirements; and, once the dataset satisfies the dataset requirements, executing a transition process (S60, S160) to proceed to model training.
3. A data collection method according to claim 2, wherein, if a dataset that satisfies the dataset requirements cannot be completed using a plurality of stored vehicle data, the method generates collection condition data indicating collection conditions for collecting the vehicle data necessary to satisfy the dataset requirements, and distributes the generated collection condition data to a plurality of in-vehicle devices.
4. A data collection method according to claim 2 or claim 3, wherein, upon acquiring the dataset requirement data, the method determines whether the dataset can be completed using a plurality of stored vehicle data, and if it cannot be completed using the plurality of stored vehicle data, the method repeats a cycle of sequentially collecting the vehicle data necessary to satisfy the dataset requirement and storing the vehicle data collected from a plurality of in-vehicle devices until the dataset is completed.
5. A data collection method according to claim 4, wherein data collection from a plurality of in-vehicle devices is accelerated when a preset acceleration condition is met during the repetition of the cycle.
6. A data collection method according to claim 5, wherein data indicating the acceleration conditions is obtained from a user via a user interface (107).
7. A data collection method according to any one of claims 1 to 3, wherein the dataset requirement includes at least one of a target scene which is the scene to be trained on the model, the number of vehicle data corresponding to the target scene, the distribution of the vehicle data corresponding to the target scene, the type of vehicle from which the vehicle data is to be collected, the period for which the vehicle data is to be collected, and preprocessing performed on the vehicle data before the model training is performed.
8. A data collection system (1) comprising: a plurality of in-vehicle devices (2) mounted on each of a plurality of vehicles and configured to transmit vehicle data including at least information about the vehicle on which it is mounted; and a center (3) configured to receive the vehicle data from the plurality of in-vehicle devices, wherein the center comprises: an acquisition unit (S10) configured to acquire dataset requirement data indicating dataset requirements that define a dataset including a plurality of the vehicle data used in model training performed on a machine learning model; and a dataset creation unit (S20 to S160) configured to collect a plurality of the vehicle data from the plurality of in-vehicle devices and to create a dataset that satisfies the dataset requirements indicated by the acquired dataset requirement data using the plurality of the vehicle data.
9. A model generation method comprising: a center (3) configured to receive vehicle data from multiple in-vehicle devices (2) installed in each of multiple vehicles, each containing at least information about the vehicle on which the device is installed; the center obtaining dataset requirement data indicating dataset requirements that define a dataset containing multiple vehicle data to be used in model training for a machine learning model; the center collecting multiple vehicle data from the multiple in-vehicle devices; using the collected multiple vehicle data to create a dataset that satisfies the dataset requirements indicated by the obtained dataset requirement data; and the center generating the machine learning model by performing model training using the dataset.