In-vehicle device, roadside device, control method, and computer program

Abstract

An in-vehicle device according to the present invention is provided in a vehicle and includes: a communication unit that receives reception target data from a roadside device, which is a device located outside of the vehicle; and a learning unit that learns a reception condition of the reception target data by evaluating a degree to which the reception target data received by the communication unit has been used by a function control device provided in the vehicle. The learning unit identifies the reception condition in a learning period in which the communication unit repeatedly receives the reception target data. After the reception condition is identified by the learning unit, the communication unit receives the reception target data according to whether or not the reception condition is satisfied.

Description

[Technical Field]

[0001] This disclosure relates to an in-vehicle device, a roadside device, a control method, and a computer program. This application claims priority to Japanese Patent Application No. 2021-200942 filed on December 10, 2021, and incorporates by reference all of the contents of that Japanese application. [Background technology]

[0002] A system for linking an on-board device mounted on an automobile, motorcycle, or the like (hereinafter referred to as a vehicle) with an external device such as a server computer (hereinafter simply referred to as a server) has been proposed. For example, data is uploaded from the on-board device to the external device via wireless communication, and the external device uses the received data in various services it provides. One service provided by the external device is one that provides information to assist the driver of the vehicle.

[0003] Modern vehicles are equipped with various electronic devices and ECUs (Electronic Control Units) that control them. For example, autonomously driven vehicles are equipped with an autonomous driving ECU. The autonomous driving ECU communicates with external devices as needed to obtain necessary information (e.g., road traffic information and dynamic driving assistance information) and controls the driving of the vehicle using the obtained information. Other ECUs include an engine control ECU, a stop-start control ECU, a transmission control ECU, an airbag control ECU, a power steering control ECU, and a hybrid control ECU. For autonomously driven vehicles, external devices provide services such as remote monitoring and remote control.

[0004] Patent Document 1 below discloses a vehicle-side device for generating map data for autonomous driving of a vehicle, which can reduce the amount of data communicated between the vehicle and a server. When uploading probe data to a server that manages map data, this vehicle-side device changes the upload frequency to a low-frequency mode based on the vehicle's driving area, weather conditions, time of day, and server instructions. The vehicle's driving area refers to whether the vehicle is driving in a low-frequency area where the upload frequency is set lower than normal. The weather conditions refer to whether the weather is bad, such as heavy rain, heavy snow, or dense fog, and the time of day refers to whether it is nighttime. The server instruction refers to the server specifying the vehicle to be responsible for transmission. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] International Publication No. 2020 / 045318 Summary of the Invention

[0006] An on-board device according to one aspect of the present disclosure is an on-board device mounted on a vehicle, and includes a communication unit that receives data to be received from a roadside device that is a device located outside the vehicle, and a learning unit that learns the reception conditions for the data to be received by evaluating the extent to which the data to be received received by the communication unit has been used by a function control device mounted on the vehicle, wherein the learning unit identifies the reception conditions during a learning period in which the communication unit repeatedly receives the data to be received, and the communication unit receives the data to be received depending on whether the reception conditions are met after the reception conditions have been identified by the learning unit. [Brief explanation of the drawings]

[0007] [Figure 1] FIG. 1 is a schematic diagram showing the configuration of a linkage system including an in-vehicle device and a server. [Figure 2] FIG. 2 is a block diagram showing a hardware configuration of the in-vehicle device shown in FIG. [Figure 3]FIG. 3 is a block diagram illustrating a hardware configuration of the in-vehicle / out-vehicle linking unit illustrated in FIG. [Figure 4] FIG. 4 is a block diagram showing a hardware configuration of the server shown in FIG. [Figure 5] FIG. 5 is a block diagram showing a schematic hierarchical structure of software in a vehicle (specifically, an on-board device) and a server. [Figure 6] FIG. 6 is a block diagram showing the functional configuration of the in-vehicle / out-vehicle linking unit related to downloading. [Figure 7] FIG. 7 is a flowchart showing a process relating to downloading executed by the vehicle interior / exterior linking unit. [Figure 8] FIG. 8 is a flowchart showing a learning process related to downloading executed by the in-vehicle / out-vehicle linking unit. [Figure 9] FIG. 9 is a block diagram showing the functional configuration of the in-vehicle / out-of-vehicle linking unit and the server regarding uploading. [Figure 10] FIG. 10 is a flowchart showing the upload process executed by the vehicle interior / exterior linking unit. [Figure 11] FIG. 11 is a flowchart showing a learning process related to uploading executed by the in-vehicle / out-of-vehicle linking unit. [Figure 12] FIG. 12 is a flowchart showing the processing executed by the server. [Figure 13] FIG. 13 is a block diagram showing the functional configuration of the in-vehicle / out-vehicle linking unit and the server according to a modified example. DETAILED DESCRIPTION OF THE INVENTION

[0008] [Problem to be solved by this disclosure] In the above-mentioned services provided by the integrated system (hereinafter also referred to as connected services), the transmission and reception of data between the in-vehicle device and the roadside device (e.g., a server) is essential. However, if data is repeatedly transmitted and received between the in-vehicle device and the server, the amount of communication data and the usage fee for the wireless line increase in proportion to the vehicle's traveling time, which may result in a shortage of wireless communication network resources. Note that "repeated" includes both regular (e.g., periodic) and irregular communication. On the other hand, simply reducing the frequency of communication between the in-vehicle device and the server will degrade the quality of the connected service.

[0009] The above problem cannot be solved by Patent Document 1. That is, while the technology disclosed in Patent Document 1 can reduce the frequency of uploads from an in-vehicle device to a server, it has the problem of not being able to reduce the amount of data that the in-vehicle device downloads from the server. In addition, it is necessary to separately clarify the predetermined conditions for reducing the upload frequency (i.e., low-frequency areas, weather conditions, time periods, etc.) through separate measurements, etc., and set them in the in-vehicle device in advance, which is cumbersome.

[0010] Therefore, an object of the present disclosure is to provide an in-vehicle device, a roadside device, a control method, and a computer program that can appropriately reduce the frequency of communication between the in-vehicle device and the roadside device without degrading the quality of connected services.

[0011] [Effects of this disclosure] According to the present disclosure, it is possible to provide an in-vehicle device, a roadside device, a control method, and a computer program that can appropriately reduce the frequency of communication between an in-vehicle device and a roadside device without degrading the quality of connected services.

[0012] [Description of the embodiments of the present disclosure] The contents of the embodiments of the present disclosure will be listed and explained below. At least some of the embodiments described below may be combined in any combination.

[0013] (1) An in-vehicle device according to a first aspect of the present disclosure is an in-vehicle device mounted on a vehicle, and includes: a communication unit that receives data to be received from a roadside device that is a device located outside the vehicle; and a learning unit that learns reception conditions for the data to be received by evaluating the extent to which the data to be received received by the communication unit has been used by a function control device mounted on the vehicle. The learning unit identifies the reception conditions during a learning period in which the communication unit repeatedly receives the data to be received, and after the reception conditions have been identified by the learning unit, the communication unit receives the data to be received depending on whether the reception conditions are met. This allows the in-vehicle device to appropriately reduce the frequency at which data for a service provided by the roadside device is downloaded by the in-vehicle device without degrading the quality of the connected service, thereby preventing unnecessary downloads.

[0014] (2) In the above (1), the learning unit may include an acquisition unit that acquires output data from the function control device and driving state data representing the driving state of the vehicle, a surrounding condition detection unit that generates surrounding condition data representing the surrounding conditions of the vehicle, an evaluation unit that generates a first evaluation index for evaluating the driving of the vehicle from the output data, and a determination unit that determines whether the target data to be received has been effectively used by the function control device by comparing the first evaluation index when the target data is received with the first evaluation index when the target data is not received during the learning period, and when the determination unit determines that the target data to be received has been effectively used, the driving state data and surrounding condition data when the target data was received can be specified as reception conditions. This makes it possible to identify conditions under which downloaded data can be effectively used, and effectively suppress wasteful downloads of data that will not be effectively used.

[0015] (3) In the above (2), the first evaluation index may include at least one of comfort, traffic efficiency, and safety related to vehicle travel, thereby making it possible to appropriately determine whether the downloaded data has been used effectively.

[0016] (4) In (2) or (3) above, the determination unit may determine whether the function control device has effectively utilized the data to be received by determining whether a difference between the first evaluation index when the data to be received and the first evaluation index when the data to be received is not received during the learning period is equal to or greater than a predetermined value greater than 0. After the learning unit has identified the reception conditions, if the difference becomes smaller than the predetermined value, the learning unit may cause the communication unit to repeatedly receive the data to be received and re-execute the process of identifying the reception conditions. As a result, if the learning results become invalid, re-learning can be quickly executed to re-determine appropriate reception conditions.

[0017] (5) In any one of (1) to (4) above, when the function control device is updated, the learning unit may cause the communication unit to repeatedly receive the data to be received and execute the process of identifying the reception conditions again. This allows appropriate reception conditions to be quickly determined again.

[0018] (6) In any one of (1) to (5) above, the communication unit may further transmit the transmission target data to the roadside device and receive from the roadside device a second evaluation index that evaluates the service provided by the roadside device, the learning unit may further learn transmission conditions for the transmission target data using the second evaluation index and identify the transmission conditions during a period in which the communication unit repeatedly transmits the transmission target data, and the communication unit may transmit the transmission target data depending on whether the transmission conditions are satisfied after the transmission conditions are identified by the learning unit. This makes it possible to appropriately reduce the frequency at which the in-vehicle device uploads data to the roadside device and suppress unnecessary uploads without degrading the quality of the connected service.

[0019] (7) An on-board device according to a second aspect of the present disclosure is an on-board device mounted on a vehicle, and includes a communication unit that receives data to be received from a roadside device that is a device located outside the vehicle, and a learning unit that learns whether the data to be received is appropriate or not. The learning unit includes an acquisition unit that acquires output data of a function control device mounted on the vehicle and driving state data that represents the driving state of the vehicle, a surrounding condition detection unit that generates surrounding condition data that represents the surrounding conditions of the vehicle, an evaluation unit that generates an evaluation index that evaluates the driving of the vehicle from the output data during a predetermined period in which the communication unit repeatedly receives the data to be received, and an evaluation unit that evaluates the driving of the vehicle during the predetermined period in which the data to be received is received. The system includes a determination unit that determines whether the data to be received has been effectively used by the function control device by comparing an evaluation index with an evaluation index when the data to be received has not been received, and a model that outputs data indicating the appropriateness of reception in accordance with input data including driving condition data and surrounding situation data, wherein the learning unit trains the model by machine learning using the training data, and the communication unit receives the data to be received in accordance with output data of the trained model, wherein the training data includes driving condition data and surrounding situation data collected over a predetermined period as input data and includes a determination result by the determination unit collected over the predetermined period as output data of the model. This makes it possible to appropriately reduce the frequency at which the in-vehicle device downloads data for services provided by roadside devices without degrading the quality of connected services, thereby preventing unnecessary downloads.

[0020] (8) According to a third aspect of the present disclosure, there is provided an on-board device mounted on a vehicle, the on-board device including: a communication unit that transmits data to be transmitted to a roadside device that is a device located outside the vehicle and receives from the roadside device an evaluation index that evaluates a service provided by the roadside device; and a learning unit that learns transmission conditions for the data to be transmitted using the evaluation index, the learning unit identifying the transmission conditions during a learning period in which the communication unit repeatedly transmits the data to be transmitted, and the communication unit transmitting the data to be transmitted depending on whether the transmission conditions are satisfied after the identification of the transmission conditions by the learning unit. This allows the on-board device to appropriately reduce the frequency at which data is uploaded to the roadside device without degrading the quality of connected services, thereby preventing unnecessary uploads.

[0021] (9) In the above (8), the communication unit can further transmit driving state data representing the driving state of the vehicle to the roadside device, and the evaluation index can be generated by the roadside device taking into account the driving state data and the surrounding conditions of the vehicle, and the learning unit can include a surrounding condition detection unit that generates surrounding condition data representing the surrounding conditions, and a determination unit that determines whether the data to be transmitted was effectively used by the roadside device by comparing the evaluation index when the data to be transmitted was transmitted with the evaluation index when the data to be transmitted was not transmitted during the learning period, and upon determining that the data to be transmitted was effectively used, the determination unit can specify the driving state data and surrounding condition data when the data to be transmitted was transmitted as transmission conditions. This allows the roadside device to appropriately generate the evaluation index, and the in-vehicle device to appropriately determine transmission conditions for uploading data to the roadside device.

[0022] (10) In the above (9), the determination unit may determine whether the roadside device has effectively used the data to be transmitted by determining whether the difference between the evaluation index when the data to be transmitted was transmitted and the evaluation index when the data to be transmitted was not transmitted during the learning period is equal to or greater than a predetermined value greater than 0. After the learning unit has identified the transmission conditions, if the difference becomes smaller than the predetermined value, the learning unit may cause the communication unit to repeatedly transmit the data to be transmitted and re-execute the process of identifying the transmission conditions. In this way, if the learning results become invalid, re-learning can be quickly executed to re-determine appropriate transmission conditions.

[0023] (11) An on-board device according to a fourth aspect of the present disclosure is an on-board device mounted on a vehicle, the on-board device including: a communication unit that transmits data to be transmitted to a roadside device that is a device located outside the vehicle, and receives from the roadside device an evaluation index that evaluates a service provided by the roadside device; and a learning unit that learns whether or not the transmission of the data to be transmitted by the communication unit is appropriate; the communication unit further transmits driving state data that represents a driving state of the vehicle to the roadside device; the roadside device generates the evaluation index by taking into account the driving state data and a surrounding situation of the vehicle; the learning unit includes a surrounding situation detection unit that generates surrounding situation data that represents the surrounding situation; and a learning unit that learns whether or not the transmission of the data to be transmitted by the communication unit is appropriate during a predetermined period in which the communication unit repeatedly transmits the data to be transmitted. and a model that outputs data indicating the appropriateness of transmission according to input data including driving condition data and surrounding situation data, wherein the learning unit trains the model by machine learning using the learning data, and the communication unit transmits the data to be transmitted according to output data of the learned model, wherein the learning data includes, as input data, the driving condition data and surrounding situation data collected over a predetermined period, and includes, as output data of the model, a determination result by the determination unit collected over the predetermined period. This makes it possible to appropriately reduce the frequency at which the in-vehicle device uploads data to the roadside device and suppress unnecessary uploads without degrading the quality of connected services.

[0024] (12) A roadside device according to a fifth aspect of the present disclosure is a roadside device that communicates with any one of the above-described in-vehicle devices (8) to (11), and includes a service execution unit that executes a predetermined service, an evaluation unit that generates an evaluation index that indicates the degree to which data to be transmitted has been used by the service execution unit, and a communication unit that transmits the evaluation index to the in-vehicle device. This allows the in-vehicle device to appropriately determine transmission conditions for uploading data to the roadside device, appropriately reduce the frequency with which the in-vehicle device uploads data to the roadside device, and suppress unnecessary uploads.

[0025] (13) A roadside device according to a sixth aspect of the present disclosure includes a communication unit that communicates with an on-board device mounted on a vehicle and receives data to be transmitted, a service execution unit that executes a predetermined service, and a learning unit that learns transmission conditions for the data to be transmitted by the on-board device by evaluating the extent to which the data to be transmitted received by the communication unit has been used by the service execution unit, wherein the learning unit identifies the transmission conditions during a learning period in which the communication unit receives the data to be transmitted repeatedly, and the communication unit transmits the transmission conditions identified by the learning unit to the on-board device. As a result, the on-board device that has received the transmission conditions can appropriately reduce the frequency at which the on-board device uploads data to the roadside device, thereby suppressing unnecessary uploads.

[0026] (14) A control method according to a seventh aspect of the present disclosure is a control method for an in-vehicle device mounted on a vehicle, the control method including: a communication step of receiving data to be received from a roadside device that is a device located outside the vehicle; and a learning step of learning reception conditions for the data to be received by evaluating the extent to which the data to be received received in the communication step has been used by a function control device mounted on the vehicle, the learning step including a step of identifying reception conditions during a learning period in which the data to be received is repeatedly received in the communication step; and a step of receiving the data to be received depending on whether the reception conditions are satisfied after the identification of the reception conditions in the learning step. This allows the in-vehicle device to appropriately reduce the frequency at which data for a service provided by the roadside device is downloaded, and prevents unnecessary downloads, without degrading the quality of connected services.

[0027] (15) A control method according to an eighth aspect of the present disclosure is a control method for an on-board device mounted on a vehicle, the control method including a communication step of receiving data to be received from a roadside device that is a device located outside the vehicle, and a learning step of learning whether the data to be received by the communication step is appropriate. The learning step includes an acquisition step of acquiring output data of a function control device mounted on the vehicle and driving state data representing a driving state of the vehicle, a surrounding situation detection step of generating surrounding situation data representing a surrounding situation of the vehicle, an evaluation step of generating an evaluation index for evaluating the driving of the vehicle from the output data during a predetermined period in which the data to be received is repeatedly received by the communication step, and an evaluation step of generating an evaluation index for evaluating the driving of the vehicle during the predetermined period in which the data to be received is received. the communication step includes a step of receiving the data to be received in response to output data of the trained model, the training data including the driving condition data and the surrounding condition data collected over a predetermined period as input data, and the training data including the determination result by the determination step collected over the predetermined period as output data of the model. This makes it possible to appropriately reduce the frequency at which the in-vehicle device downloads data for services provided by the roadside device without degrading the quality of the connected service, thereby suppressing unnecessary downloads.

[0028] (16) A control method according to a ninth aspect of the present disclosure is a control method for an on-board device mounted on a vehicle, the control method including: a communication step of transmitting data to be transmitted to a roadside device that is a device located outside the vehicle and receiving from the roadside device an evaluation index indicating the extent to which the data to be transmitted has been used by the roadside device; and a learning step of learning transmission conditions for the data to be transmitted using the evaluation index, the learning step including a step of identifying the transmission conditions during a learning period in which the data to be transmitted is repeatedly transmitted by the communication step, and the communication step including a step of transmitting the data to be transmitted depending on whether the transmission conditions are satisfied after the identification of the transmission conditions by the learning step. This makes it possible to appropriately reduce the frequency at which the on-board device uploads data to the roadside device and suppress unnecessary uploads without degrading the quality of connected services.

[0029] (17) A control method according to a tenth aspect of the present disclosure is a control method for an on-board device mounted on a vehicle, the control method including a communication step of transmitting data to be transmitted to a roadside device that is a device located outside the vehicle and receiving an evaluation index from the roadside device that evaluates a service provided by the roadside device, and a learning step of learning whether or not the transmission of the data to be transmitted by the communication step is appropriate, the communication step including a step of transmitting driving state data that represents a driving state of the vehicle to the roadside device, the evaluation index being generated by the roadside device taking into consideration the driving state data and a surrounding situation of the vehicle, the learning step including a surrounding situation detection step of generating surrounding situation data that represents the surrounding situation, and a learning step of learning whether or not the transmission of the data to be transmitted by the communication step is appropriate during a predetermined period of time in which the data to be transmitted is repeatedly transmitted by the communication step. the communication step further includes a step of transmitting the data to be transmitted in accordance with output data of the trained model, the training data including, as input data, the driving condition data and the surrounding condition data collected over a predetermined period, and the training data including, as output data of the model, a determination result obtained by the determination step and collected over the predetermined period. This makes it possible to appropriately reduce the frequency of data uploads from the in-vehicle device to the roadside device and suppress unnecessary uploads, without degrading the quality of connected services.

[0030] (18) A computer program according to an eleventh aspect of the present disclosure causes a computer mounted on a vehicle to implement a communication function for receiving data to be received from a roadside device located outside the vehicle, and a learning function for learning reception conditions for the data to be received by evaluating the extent to which the data to be received received by the communication function has been used by a function control device mounted on the vehicle, the learning function including a function for identifying reception conditions during a learning period during which the data to be received is repeatedly received by the communication function, and the communication function including a function for receiving the data to be received depending on whether the reception conditions are met after the identification of the reception conditions by the learning function. This allows the frequency at which the in-vehicle device downloads data for a service provided by the roadside device to be appropriately reduced without degrading the quality of the connected service, thereby preventing unnecessary downloads.

[0031] (19) A computer program according to a twelfth aspect of the present disclosure causes a computer mounted on a vehicle to realize a communication function for receiving data to be received from a roadside device that is a device located outside the vehicle, and a learning function for learning whether or not the data to be received is appropriate, wherein the learning function includes an acquisition function for acquiring output data from a function control device mounted on the vehicle and driving state data representing the driving state of the vehicle, a surrounding situation detection function for generating surrounding situation data representing the surrounding situation of the vehicle, an evaluation function for generating an evaluation index for evaluating the driving of the vehicle from the output data during a predetermined period in which the data to be received is repeatedly received by the communication function, and a learning function for learning whether or not the data to be received is appropriate. The system includes a determination function that determines whether the data to be received was effectively used by the function control device by comparing an evaluation index when the data to be received with an evaluation index when the data to be received was not received, and a function that uses learning data to train a model that outputs data indicating the appropriateness of reception in response to input data including driving condition data and surrounding situation data, the communication function including a function that receives the data to be received in response to output data of the learned model, the learning data including driving condition data and surrounding situation data collected over a predetermined period as input data, and the determination result by the determination function collected over the predetermined period as output data of the model. This makes it possible to appropriately reduce the frequency at which the in-vehicle device downloads data for services provided by roadside devices without degrading the quality of connected services, thereby preventing unnecessary downloads.

[0032] (20) A computer program according to a thirteenth aspect of the present disclosure provides a computer program for implementing, in a vehicle, a communication function for transmitting data to be transmitted to a roadside device that is a device located outside the vehicle and receiving, from the roadside device, an evaluation index indicating the extent to which the data to be transmitted has been used by the roadside device, and a learning function for learning transmission conditions for the data to be transmitted using the evaluation index, wherein the learning function includes a function for identifying the transmission conditions during a learning period in which the communication function repeatedly transmits the data to be transmitted, and the communication function includes a function for transmitting the data to be transmitted depending on whether the transmission conditions are satisfied after the identification of the transmission conditions by the learning function. This allows the frequency at which the in-vehicle device uploads data to the roadside device to be appropriately reduced and prevents unnecessary uploads without degrading the quality of connected services.

[0033] (21) A computer program according to a fourteenth aspect of the present disclosure provides a computer mounted on a vehicle with a communication function for transmitting data to be transmitted to a roadside device, which is a device located outside the vehicle, and receiving from the roadside device an evaluation index that evaluates a service provided by the roadside device, and a learning function for learning whether or not the transmission of the data to be transmitted by the communication function is appropriate, wherein the communication function includes a function for transmitting driving state data that represents the driving state of the vehicle to the roadside device, and the evaluation index is generated by the roadside device taking into consideration the driving state data and the surrounding conditions of the vehicle, and the learning function includes a surrounding condition detection function for generating surrounding condition data that represents the surrounding conditions, and a learning function for repeatedly transmitting the data to be transmitted by the communication function during a predetermined period. and a function for performing machine learning using the training data to create a model that outputs data indicating the appropriateness of transmission in response to input data including driving condition data and surrounding situation data, wherein the communication function further includes a function for transmitting the data to be transmitted in response to output data of the trained model, wherein the training data includes driving condition data and surrounding situation data collected over a predetermined period as input data, and includes a determination result by the determination function collected over the predetermined period as output data of the model. This makes it possible to appropriately reduce the frequency at which the in-vehicle device uploads data to the roadside device without degrading the quality of the connected service, thereby suppressing unnecessary uploads.

[0034] [Details of the embodiments of the present disclosure] In the following embodiments, the same components are denoted by the same reference numerals, and their names and functions are also the same, so detailed descriptions thereof will not be repeated.

[0035] [Overall configuration] Referring to FIG. 1 , the cooperation system according to the embodiment of the present disclosure includes an on-board device 100 and an on-board device 110 mounted on a vehicle 102 and a vehicle 112, respectively, a base station 104, and a server 106. The base station 104 is connected to a network 108 such as the Internet. The base station 104 is a base station for wireless communication such as cellular communication. The base station 104 provides mobile communication services using, for example, LTE (Long Term Evolution), 4G (Fourth Generation Mobile Communication System) lines, and 5G (Fifth Generation Mobile Communication System) lines. The on-board device 100 and the on-board device 110 can communicate with the server 106 via the base station 104 and the network 108. The base station 104 may provide wireless communication functions such as Wi-Fi and C-V2X (Cellular-Vehicle to Everything).

[0036] Each of the vehicles 102 and 112 is equipped with a sensor such as an image sensor, and the sensor data output from the sensor is acquired by the in-vehicle device 100 and the in-vehicle device 110 and transmitted (uploaded) to the server 106. The server 106 performs services such as providing driving assistance information, remote monitoring, and remote control, and uses the sensor data received from the in-vehicle device 100 and the in-vehicle device 110 for the services it provides. The server 106 may be a device installed outside the vehicle 102, the vehicle 112, etc., or may be a roadside device fixedly installed on a road or its surroundings.

[0037] The infrastructure sensor 114 is a device equipped with a sensor function that is installed on a road or its surrounding area, and has a function for communicating with the base station 104. The infrastructure sensor 114 is, for example, an image sensor (such as a digital surveillance camera), a radar (such as a millimeter-wave radar), or a laser sensor (such as a LiDAR (Light Detection and Ranging)). The infrastructure sensor 114 has a wireless communication function such as Wi-Fi or C-V2X, and may communicate directly with the on-vehicle device 100 and the on-vehicle device 110. The infrastructure sensor 114 transmits sensor data (such as video data) to the server 106 and the on-vehicle device 100 and the on-vehicle device 110 via the base station 104 or directly.

[0038] FIG. 1 exemplarily shows one base station 104, one infrastructure sensor 114, and two vehicles 102 and 112 equipped with on-board devices. However, this is merely an example. Typically, multiple base stations and multiple infrastructure sensors are provided, and on-board devices are installed in three or more vehicles. There may be vehicles that do not have on-board devices. Vehicles that do not have on-board devices are subject to detection by sensors and infrastructure sensors installed in vehicles equipped with on-board devices.

[0039] [Hardware configuration of on-board device] Referring to FIG. 2, an example of the hardware configuration of an in-vehicle device 100 mounted on a vehicle 102 is shown. The in-vehicle device 110 mounted on a vehicle 112 is similarly configured. The in-vehicle device 100 includes a communication unit 120 and an in-vehicle / out-of-vehicle linking unit 122. FIG. 2 shows the communication unit 120, sensors 124, an autonomous driving ECU 126, a drive ECU 128, and a bus 130 mounted on the vehicle 102. The vehicle 102 is equipped with a plurality of ECUs as devices for controlling various functions of the vehicle (i.e., function control devices). FIG. 2 representatively shows the autonomous driving ECU 126 and the drive ECU 128.

[0040] The communication unit 120 performs wireless communication with devices outside the vehicle 102 via the base station 104. The communication unit 120 includes an integrated circuit (IC) for performing modulation and multiplexing employed in the wireless communication service provided by the base station 104, an antenna for transmitting and receiving radio waves at a predetermined frequency, an RF circuit, and the like. The communication unit 120 also has a function for communicating with a global navigation satellite system (GNSS) such as a global positioning system (GPS). The communication unit 120 also has a wireless communication function such as Wi-Fi or C-V2X, and may communicate directly with the infrastructure sensor 114.

[0041] The in-vehicle / external linking unit 122 is responsible for connecting the communication function with the outside of the vehicle (i.e., communication specifications) with the communication function within the vehicle (i.e., communication specifications) (i.e., communication protocol conversion, etc.). The autonomous driving ECU 126 can communicate with external devices via the in-vehicle / external linking unit 122 and the communication unit 120. The in-vehicle / external linking unit 122 receives driving assistance information from the server 106 via the communication unit 120 and transfers it to the autonomous driving ECU 126. The in-vehicle / external linking unit 122 also receives sensor data from the infrastructure sensor 114 via the communication unit 120. The bus 130 is responsible for communication functions within the vehicle. Communication (i.e., data exchange) between the in-vehicle / external linking unit 122, the sensor 124, the autonomous driving ECU 126, and the drive ECU 128 is performed via the bus 130. For example, Ethernet (registered trademark), CAN (Control Area Network), etc. are used for the bus 130.

[0042] The sensor 124 is mounted on the vehicle 102 and includes sensors for acquiring information about the inside and outside of the vehicle 102. Sensors for acquiring information about the outside of the vehicle include imaging devices (for example, digital cameras (CCD cameras, CMOS cameras)), radar, laser sensors, etc. Sensors for acquiring information about the inside of the vehicle include imaging devices. The sensor 124 acquires information within a detection range (for example, an imaging range in the case of a camera) and outputs it as sensor data. In the case of a digital camera, it outputs digital video data. The detection signal (analog or digital) of the sensor 124 is output as digital data to the bus 130 via an interface unit (not shown), and is transmitted to the vehicle interior / exterior linking unit 122, the autonomous driving ECU 126, etc.

[0043] The autonomous driving ECU 126 controls the driving of the vehicle 102. For example, the autonomous driving ECU 126 acquires sensor data, analyzes it to understand the situation around the vehicle, and generates control information (e.g., information on acceleration (deceleration), speed, driving direction, etc.) taking into account the current driving state of the vehicle 102 (e.g., position, speed, etc.), and outputs the control information to the driving ECU 128. The driving ECU 128 controls mechanisms related to autonomous driving (i.e., mechanisms such as the engine, transmission, steering, and brakes) using the control information input from the autonomous driving ECU 126. The autonomous driving ECU 126 can use driving assistance information (i.e., dynamic information, etc.) acquired from the in-vehicle / external communication unit 122 to generate the control information. The autonomous driving ECU 126 acquires information representing the current driving state of the vehicle 102 (e.g., position, speed, etc.) from the GPS and the driving ECU 128. For example, the position information can be generated from GPS data, and the speed information can be acquired from the autonomous driving ECU 126.

[0044] 3, the vehicle interior / exterior linking unit 122 includes a control unit 140, a memory 142, an I / F unit 144, and a learning unit 146. The control unit 140 includes a CPU (Central Processing Unit) and controls the memory 142, the I / F unit 144, and the communication unit 120 (control signals are indicated by dotted arrows). The memory 142 is, for example, a rewritable nonvolatile semiconductor memory, and stores a computer program (hereinafter simply referred to as a program) executed by the control unit 140. The memory 142 provides a work area for the program executed by the control unit 140. The memory 142 also stores data received from the outside via the communication unit 120 (sensor data from the infrastructure sensor 114, downloaded data from the server 106, etc.).

[0045] Under the control of the control unit 140, the I / F unit 144 serves as an interface with the sensor 124, the autonomous driving ECU 126, and the drive ECU 128 (see FIG. 2). Data acquired by the I / F unit 144 from the sensor 124, the autonomous driving ECU 126, and the drive ECU 128 is stored in the memory 142. The data acquired from the sensor 124 (i.e., sensor data) is data to be transmitted to the server 106. The communication unit 120 is controlled by the control unit 140 to generate and transmit packet data from the upload data stored in the memory 142. If the sensor data is video data, a predetermined amount of data sequentially acquired from the sensor 124 is buffered in the memory 142, and transmission data is generated from the buffered data and transmitted sequentially.

[0046] As will be described later, the learning unit 146 learns the conditions for downloading data from the server 106 (i.e., reception conditions) and the conditions for uploading data to the server 106 (i.e., transmission conditions) in order to appropriately reduce the communication frequency. The learning results are stored in the memory 142. Various data are downloaded from the server 106. The download data subject to the reception conditions (hereinafter also referred to as data to be received) is data for a service provided by the server 106 (i.e., driving assistance information, etc.). In addition, various data is uploaded to the server 106. The upload data subject to the transmission conditions (hereinafter also referred to as data to be transmitted) is data that can be used by a service provided by the server 106 (i.e., sensor data acquired from the sensor 124, etc.).

[0047] [Server hardware configuration] Referring to FIG. 4, the server 106 includes a control unit 160, a memory 162, a communication unit 164, and a bus 166. Data transfer between the units is performed via the bus 166. The control unit 160 includes a CPU, controls the units, and provides various services. The communication unit 164 receives information (i.e., sensor data, vehicle information, etc.) uploaded from the in-vehicle device 100 and the in-vehicle device 110 via the base station 104 and the network 108. The memory 162 includes a rewritable nonvolatile semiconductor memory and a large-capacity storage device such as an HDD (Hard Disk Drive). The data received by the communication unit 164 is transferred to and stored in the memory 162. The uploaded data stored in the memory 162 is used by an application program (hereinafter simply referred to as an application) for providing the service. The communication unit 164 has a function of accessing the network 108 wirelessly or via a wired connection. Under the control of the control unit 160, the communication unit 164 reads download data (i.e., service data, etc.) from the memory 162 and transmits it to the vehicles 102 and 112. The server 106 also includes an operation unit (not shown) such as a computer keyboard and mouse, which allows an administrator or the like to input instructions to the control unit 160.

[0048] [Software configuration] Referring to FIG. 5, the software of the collaboration system configured by the vehicle 102 and the server 106 is configured hierarchically. The hierarchical structure shown in FIG. 5 corresponds, for example, to the hierarchical structure of the OSI (Open Systems Interconnection) reference model. The vehicle 102 includes multiple ECUs as described above, and the upper layers include application programs (i.e., the 1-ECU application to the M-ECU application) for realizing the functions of each ECU. The applications in the upper layers are executed in parallel by microcomputers or the like installed in each ECU. The upper layers correspond, for example, to the application layer of the OSI reference model. The vehicle 102 includes, in its lower layers, a communication stack (e.g., the session layer and below of the OSI reference model) that handles communication with the outside world, and in its middle layers, a sublayer program that mediates between the programs in the upper layers and the programs in the lower layers. The middle layer corresponds, for example, to the presentation layer of the OSI reference model.

[0049] The functions (or corresponding programs) of the in-vehicle-external linking unit 122 of the in-vehicle device 100 are mainly positioned as sub-layer programs. The in-vehicle-external linking unit 122 controls the communication stack of the lower layer and transmits data to be transmitted (i.e., sensor data acquired from the sensor 124, etc.) to the server 106, as described above. The in-vehicle-external linking unit 122 also controls the communication stack of the lower layer and, as described above, receives data to be received (i.e., driving assistance information, etc.) from the server 106 and passes the data to, for example, the autonomous driving ECU 126 (see FIG. 2 ). The in-vehicle-external linking unit 122 controls the communication stack of the lower layer and, as described below, receives evaluation indexes from the server 106 to be used for identifying transmission conditions for data to be transmitted. The functions (or corresponding programs) of the in-vehicle-external linking unit 122 may include some of multiple applications in the upper layer. That is, the in-vehicle-external linking unit 122 may include the function of an ECU.

[0050] As described above, the server 106 executes a plurality of services (i.e., driving assistance, remote monitoring, remote control, etc.), and includes application programs (i.e., first service application to Nth service application) for realizing each service in its upper layer. The server 106 includes a communication stack in its lower layer that handles communication with the outside (i.e., the in-vehicle device 100 of the vehicle 102, etc.), and includes a sublayer program in its middle layer that mediates between the programs in the upper layer and the programs in the lower layer. The upper layer, middle layer, and lower layer correspond to, for example, the application layer, presentation layer, and session layer and lower layers of the OSI reference model, respectively.

[0051] A program for implementing an infrastructure interworking unit that communicates with the in-vehicle device 100 of the vehicle 102 to configure an interworking system is positioned as a sub-layer program. The upper-layer and middle-layer programs are executed as multitasks by the control unit 160 (see FIG. 4). The infrastructure interworking unit controls the lower-layer communication stack and transmits service data (i.e., driving support information, etc.) provided by the upper-layer program to the in-vehicle device 100 of the vehicle 102. The infrastructure interworking unit also controls the lower-layer communication stack and, as described above, receives sensor data, etc. uploaded from the vehicle 102 and passes it to the upper-layer program. The infrastructure interworking unit observes the operation of the upper-layer program and generates an evaluation index that indicates how effectively the sensor data, etc. passed to the upper-layer program is used by the upper-layer program. The server 106 controls the lower-layer communication stack and transmits the evaluation index to the corresponding in-vehicle device.

[0052] [Download Functionality] The download function of the vehicle interior / exterior linkage unit 122 will be described with reference to Fig. 6. As shown in Fig. 4, the vehicle interior / exterior linkage unit 122 includes a control unit 140, a memory 142, an I / F unit 144, and a learning unit 146. The learning unit 146 includes a surrounding condition detection unit 200, an evaluation unit 202, and a determination unit 204.

[0053] The memory 142 stores sensor data 210, downloaded data 212, driving condition data 214, surrounding condition data 216, evaluation result data 218, and learning result data 220. The sensor data 210 includes sensor data output from the sensor 124 and sensor data received from the infrastructure sensor 114 via the communication unit 120. The downloaded data 212 includes reception target data received from the server 106 via the communication unit 120. The driving condition data 214 is data representing the driving condition (e.g., position, speed, etc.) of the vehicle 102. The driving condition data is acquired from the autonomous driving ECU 126 by the I / F unit 144. The surrounding condition data 216 is data representing traffic conditions around the vehicle 102 (i.e., traffic accidents, traffic congestion, etc.) and is detected by the surrounding condition detection unit 200, as described below. The evaluation result data 218 is data representing how effectively the downloaded data was used by the autonomous driving ECU 126 and is generated by the evaluation unit 202, as described below. The learning result data 220 includes conditions for executing communication with the server 106 (i.e., reception conditions), and is identified by the determination unit 204, as described below. The sensor data 210, the download data 212, the driving condition data 214, the surrounding condition data 216, and the evaluation result data 218 are stored with corresponding time information (e.g., a timestamp when stored in the memory 142). Time information is not necessarily assigned to each piece of data; one piece of time information (e.g., a representative timestamp) may be assigned to a group of data stored close in time. By storing the data with time information, it is possible to identify data (or groups of data) that correspond to each other using the time information.

[0054] The I / F unit 144 is controlled by the control unit 140, acquires sensor data output from the sensor 124, and stores the sensor data in the memory 142 as sensor data 210. The communication unit 120 stores sensor data received from the infrastructure sensor 114 in the memory 142 as sensor data 210. The I / F unit 144 acquires data representing the driving state (e.g., position, speed, etc.) of the vehicle 102 from the autonomous driving ECU 126, and stores the data in the memory 142 as driving state data 214.

[0055] The surrounding situation detection unit 200 reads the sensor data 210 from the memory 142, detects the surrounding situation of the vehicle (i.e., the vehicle 102), and stores the detection result in the memory 142 as surrounding situation data 216. The surrounding situation data is data (e.g., data indicating the type of situation and position coordinates indicating the area where the situation occurs) that indicates the presence of a traffic accident, traffic congestion, blind spot (e.g., an area where the view cannot be seen, such as an intersection), etc. If the sensor data 210 is image data (e.g., video image data, etc.), the surrounding situation detection unit 200 can extract objects (e.g., dynamic objects such as vehicles and static objects such as buildings and road signs) by image processing and generate surrounding situation data.

[0056] The evaluation unit 202 acquires output data (i.e., control information) from the autonomous driving ECU 126 to the drive ECU 128, generates evaluation indexes, and stores them in the memory 142 as evaluation result data 218. The output data of the autonomous driving ECU 126 depends on the driving state of the vehicle 102 and the surrounding conditions of the vehicle 102. Furthermore, if download data 212 (i.e., driving assistance information, etc.) has been received, the autonomous driving ECU 126 may also use the download data 212 to generate output data, and therefore the output data of the autonomous driving ECU 126 also depends on the download data 212. If the output data of the autonomous driving ECU 126 when not using the download data 212 is Y1 and the output data when using the download data 212 is Y2, then Y1 = f(X1) and Y2 = f(X1, X2) can be expressed using a predetermined function (specifically, an algorithm) f. X1 represents a set of driving state data and surrounding condition data, and X2 represents the download data 212. For example, if blind spot information (e.g., information about a dynamic object that cannot be acquired by the sensors 124 of the vehicle 102) can be acquired as X2, the output data of the autonomous driving ECU 126 (e.g., control information of the drive ECU 128) will change depending on whether the autonomous driving ECU 126 uses the blind spot information. Note that the output data of the autonomous driving ECU 126 is usually made up of multiple parameters (i.e., parameters such as acceleration (deceleration), speed, and driving direction), and each of Y1 and Y2 is treated as a vector.

[0057] The evaluation unit 202 generates an evaluation index for the output data of the autonomous driving ECU 126 from Y1 and Y2. Because the drive ECU 128 is controlled by the output data of the autonomous driving ECU 126, this evaluation index evaluates the driving performance of the host vehicle. If the evaluation index when the downloaded data 212 is not used is Z1 and the evaluation index when the downloaded data 212 is used is Z2, then Z1 = g(Y1) and Z2 = g(Y2) can be expressed using a predetermined function (specifically, a model) g. The evaluation index is, for example, any one of safety, comfort, traffic efficiency, etc., or any combination thereof. Safety is, for example, represented (i.e., quantified) by the number of sudden stops, sudden decelerations, etc., identified from the time-series output data of the autonomous driving ECU 126. Comfort is, for example, the ride comfort of the vehicle, and is quantified by the vibrations (i.e., amplitude and frequency) of the host vehicle, etc., identified from the time-series output data of the autonomous driving ECU 126. The traffic efficiency is, for example, a link travel time (for example, a travel time for a predetermined distance), and can be calculated from the time-series travel state data 214, for example.

[0058] Under the control of the control unit 140, the determination unit 204 reads the evaluation indexes Z1 and Z2 from the evaluation result data 218 and determines whether there is a significant difference between Z1 and Z2. Specifically, if the function g is set so that the more effectively the downloaded data 212 is used, the larger the evaluation index, the determination unit 204 calculates the difference ΔZ between Z1 and Z2 (i.e., ΔZ = Z2 - Z1) and determines whether the difference ΔZ is greater than a predetermined threshold value Th. If the function g is set so that the more effectively the downloaded data 212 is used, the determination unit 204 calculates the difference ΔZ by ΔZ = Z1 - Z2. In either case, the difference is the difference from the evaluation index Z1 when the downloaded data 212 is not used (a difference (≧0) based on the evaluation index Z1). If ΔZ > Th, data corresponding to Z2 is read from the driving condition data 214 and the surrounding situation data 216 and stored in the memory 142 as the learning result data 220. If the evaluation index when the downloaded data 212 is used is more than Th away from the evaluation index when the downloaded data 212 is not used, it can be said that the downloaded data 212 was used effectively by the autonomous driving ECU 126. Therefore, if a similar state is detected thereafter, the data to be received can be downloaded from the server 106. On the other hand, if the evaluation index when the downloaded data 212 is used is Th or less away from the evaluation index when the downloaded data 212 is not used, it cannot be said that the downloaded data 212 was used effectively by the autonomous driving ECU 126. In other words, it can be said that downloading performed in such a state is wasteful.

[0059] During a predetermined period (hereinafter referred to as the learning period), the control unit 140 controls the communication unit 120 to repeatedly (e.g., periodically) download the data to be received. The download is performed by sending a predetermined request (hereinafter referred to as a download request) from the control unit 140 to the server 106. The surrounding condition detection unit 200, the evaluation unit 202, and the determination unit 204 each perform processing at a predetermined timing. This accumulates the learning result data 220. After the learning period has elapsed, the control unit 140 reads the latest driving state data 214 and surrounding condition data 216 and determines whether the read data satisfies the reception conditions, i.e., whether the read data corresponds to the learning result data 220. If the reception conditions are satisfied, the control unit 140 controls the communication unit 120 to send a download request to the server 106 and download the data to be received. If the reception conditions are not satisfied, the control unit 140 does not perform the download. Therefore, the in-vehicle / out-of-vehicle linking unit 122 downloads the data to be received (i.e., data for the connected service provided by the server 106, such as driving assistance information) when there is a high possibility that the data will be used effectively, and does not download the data in other cases. That is, the frequency with which the in-vehicle device 100 downloads data for the service provided by the server 106 (i.e., a roadside device, etc.) can be appropriately reduced without degrading the quality of the connected service, and unnecessary downloads can be suppressed.

[0060] As described above, the evaluation index may include at least one of comfort, traffic efficiency, and safety, thereby making it possible to appropriately determine whether the downloaded data has been used effectively.

[0061] [Operation of the vehicle interior / exterior linkage unit (download)] 7 and 8, the operation of the in-vehicle / out-vehicle linking unit 122 regarding downloading will be described with reference to the functions shown in FIG. 6. The process shown in FIG. 7 is implemented by the control unit 140 reading and executing a predetermined program from the memory 142. The process shown in FIG. 7 starts, for example, when the in-vehicle device 100 is powered on. The learning period T, the download cycle ΔT1 during the learning period, the learning cycle ΔT2 (e.g., ΔT2>ΔT1), the threshold value Th, and the like are assumed to be pre-stored in the memory 142. The sensor data 210 and the download data 212 are periodically stored in the memory 142 with timestamps. The surrounding condition detection unit 200 and the evaluation unit 202 each periodically execute the above-described process and store the surrounding condition data 216 and the evaluation result data 218 with timestamps. Typically, the sensor data 210, the download data 212, the surrounding condition data 216, and the evaluation result data 218 are each deleted, starting with the oldest data, until a predetermined amount is stored.

[0062] In step 300, the control unit 140 determines whether or not to perform a download. Specifically, the control unit 140 determines whether or not a period ΔT1 has elapsed since the previous download was performed. If it is determined that the period ΔT1 has elapsed, the control proceeds to step 302. If not, the control proceeds to step 304. Note that when step 300 is performed for the first time, it is determined that a download is to be performed. The control unit 140 may obtain the current time from, for example, a timer installed in the in-vehicle device 100.

[0063] In step 302, the control unit 140 controls the communication unit 120 to download the data to be received from the server 106. Specifically, the control unit 140 transmits a download request to the server 106. Thereafter, control proceeds to step 304. The server 106 that has received the download request identifies the source address of the packet data, thereby identifying the vehicle 102 (specifically, the in-vehicle device 100) that transmitted the download request, and transmits the data to be received.

[0064] In step 304, the control unit 140 determines whether or not to perform learning. Specifically, the control unit 140 determines whether or not a period ΔT2 has elapsed since the previous learning was performed. If it is determined that the period ΔT2 has elapsed, the control proceeds to step 306. If not, the control proceeds to step 318. Note that when step 304 is executed for the first time, it is determined that learning is to be performed.

[0065] In step 306, the control unit 140 learns the download conditions (i.e., the reception conditions). Specifically, the control unit 140 executes the process shown in FIG.

[0066] Referring to FIG. 8, in step 330, the control unit 140 reads the evaluation result data 218 stored in the memory 142 and calculates the difference ΔZ (≧0) between the evaluation index Z2 when the data to be received was downloaded and the evaluation index Z1 when the data was not downloaded, as described above. The evaluation index Z2 that was the processing target is the evaluation index Z2 that had not been processed up to that point. The evaluation indexes Z2 and Z1 used to calculate the difference are, for example, the evaluation indexes Z2 and Z1 that are assigned with the closest time information to each other. If the number of evaluation indexes Z2 is smaller than the number of evaluation indexes Z1, the representative value (e.g., average value) of the evaluation index Z1 that is located between two adjacent evaluation indexes Z2 when arranged in chronological order and one of the two adjacent evaluation indexes Z2 can be used to calculate the difference ΔZ. The control unit 140 stores the calculated difference ΔZ in the memory 142 with time information. For this time information, for example, the time information assigned to the evaluation index Z2 used to calculate the difference can be used. Thereafter, control proceeds to step 332.

[0067] In step 332, the control unit 140 determines whether the difference ΔZ calculated in step 330 is greater than the threshold value Th. If ΔZ>Th, the control proceeds to step 334. If not, the control proceeds to step 336.

[0068] In step 334, the control unit 140 identifies the driving condition data and surrounding condition data used to calculate the evaluation index Z2 for which it was determined in step 332 that ΔZ > Th (i.e., the evaluation index at the time of download), and stores these as learning result data 220. To identify the driving condition data and surrounding condition data, it is sufficient to compare the time information attached to them with the time information attached to the corresponding evaluation index Z2. The control unit 140 identifies the driving condition data and surrounding condition data attached with time information that is the same time as the time information attached to the evaluation index Z2, or that is close within a predetermined range.

[0069] In step 336, the control unit 140 determines whether step 332 has been executed for all of the differences ΔZ calculated in the previously executed step 330. If it is determined that all of them have been executed, the control proceeds to step 308 in Fig. 7. If not, the control returns to step 332.

[0070] Returning to FIG. 7 , in step 308, control unit 140 determines whether learning is complete. Specifically, control unit 140 determines whether learning period T has elapsed. If it is determined that learning is complete, control proceeds to step 310. If not, control proceeds to step 318.

[0071] In step 310, the control unit 140 reads the latest driving condition data 214 and surrounding condition data 216 from the memory 142 and determines whether the download conditions (i.e., reception conditions) are met. Specifically, the control unit 140 determines whether a set of values ​​identical to the set of driving condition data 214 and surrounding condition data 216 that have been read, or a set of values ​​close within a predetermined range, is stored in the learning result data 220. If it is determined that the values ​​are stored, control proceeds to step 312. Otherwise, control proceeds to step 314.

[0072] In step 312, control unit 140 executes the download, similar to step 302. That is, control unit 140 transmits a download request to server 106 and receives the data to be received transmitted from server 106. Thereafter, control proceeds to step 316.

[0073] If the determination result in step 310 is NO, in step 314, the control unit 140 determines whether or not an instruction to end the program has been received. If it is determined that an instruction to end the program has been received, the program ends. If not, the control returns to step 310. The instruction to end the program is issued, for example, by turning off the power of the in-vehicle device 100.

[0074] In step 316, the control unit 140 determines whether the current learning result is valid. Specifically, the control unit 140 reads from the memory 142 the evaluation index Z2 (i.e., the latest evaluation result data 218) representing the results of the use by the autonomous driving ECU 126 of the data to be received downloaded in the immediately preceding step 312, calculates the difference ΔZ in the same manner as above, and compares it with the threshold value Th. The evaluation index Z1 used as the reference for calculating the difference may be a representative value (e.g., average, median, etc.) of the evaluation index Z1 during the learning period. If ΔZ>Th, the current learning result (i.e., the learning result data 220 stored in the memory 142) is determined to be valid, and control returns to step 310. Otherwise, the control unit 140 deletes the learning result data 220 stored in the memory 142, and control proceeds to step 318.

[0075] In step 318, the control unit 140 determines whether an end instruction has been received, similar to step 314. If it is determined that an end instruction has been received, the program ends. If not, control returns to step 300.

[0076] As described above, by repeating the processes from step 300 to step 308, periodic downloads and learning of download conditions (reception conditions) are performed until learning is completed (the learning period has elapsed). Note that, although downloads and learning are performed periodically in the above description, this is not limiting. The frequency of repeated downloads needs to be greater than the frequency of repeated learning. After learning is completed, by repeating the processes from step 310 to step 316, it is determined whether to perform downloads using the learning results, and downloads are performed only when the download conditions are satisfied. Therefore, the frequency at which the in-vehicle device 100 downloads data for services provided by the server 106 (i.e., roadside device, etc.) can be appropriately reduced without degrading the quality of connected services, and unnecessary downloads can be suppressed.

[0077] Furthermore, if the learning result data 220 stored in memory 142 becomes invalid (i.e., if there is no longer a significant difference between evaluation index Z2 and evaluation index Z1 (specifically, the judgment result in step 316 is NO)), re-learning is performed. That is, control returns to step 300, and periodic downloading and learning of the download conditions (i.e., reception conditions) are performed again. As a result, if the learning result becomes invalid, re-learning is performed promptly, and appropriate reception conditions can be determined again.

[0078] In the above, a case has been described in which the validity of the learning results is determined and re-learning is performed if the learning results are no longer valid, but this is not limiting. Re-learning may also be performed intentionally without determining the validity of the learning results. For example, if the autonomous driving ECU 126 is updated, the output data of the autonomous driving ECU 126 may change even under the same driving conditions and surrounding circumstances. Therefore, even in such a case, it is preferable to quickly perform re-learning without using the learning results used before the update. This allows appropriate reception conditions to be quickly re-determined. Note that updating the autonomous driving ECU 126 is not limited to updating (i.e., updating) the program of the autonomous driving ECU 126, but also includes replacing it with a new type of hardware.

[0079] [Functional configuration for uploading] With reference to FIG. 9, the functions of the in-vehicle / out-of-vehicle linking unit 122 and the server 106 related to uploading will be described. The in-vehicle / out-of-vehicle linking unit 122 is configured as shown in FIG. 6. In FIG. 9, the same configuration as in FIG. 6 is shown for the in-vehicle / out-of-vehicle linking unit 122, with the configuration not used during uploading (i.e., the evaluation unit 202 and the download data 212) indicated by dashed lines. Just as an evaluation index indicating the degree to which data to be received (i.e., download data) is effectively used during downloading is introduced, an evaluation index indicating the degree to which data to be transmitted (i.e., upload data) is effectively used during uploading is used. As will be described later, the server 106 is responsible for calculating the evaluation index related to uploading, and therefore the server 106 in FIG. 9 shows the configuration required for this. Note that the "evaluation result data" stored in the memory 142 is different from the evaluation result data 218 shown in FIG. 6 and is acquired from the server 106, and is therefore denoted by the reference numeral "232" different from that in FIG. 6. In the following, redundant explanations of the same functions as during downloading will not be repeated, and differences will be mainly described.

[0080] 6, the memory 142 stores upload data 230. The upload data 230 is data to be uploaded and is, for example, data (e.g., video data) from the sensor data 210 (i.e., output data of the sensor 124) that can be effectively used by the server 106. The data to be transmitted is accompanied by information such as location information at which the data was acquired. For example, when video data acquired by an in-vehicle camera or the like is uploaded as data to be transmitted, information indicating the location and imaging direction at which the video data was acquired (i.e., captured) is added, so that the server 106 can effectively use the results of analyzing the uploaded data for generating service data, etc. The evaluation result data 232 is calculated by the evaluation unit 244 of the server 106 and received by the communication unit 120, as will be described later.

[0081] The server 106 includes a service execution unit 240, a surrounding condition detection unit 242, and an evaluation unit 244 to handle uploads from the vehicle interior / exterior cooperation unit 122. The service execution unit 240 has a function of executing connected services provided by the server 106 (for example, providing driving assistance information, remote monitoring, remote control, etc.). The service execution unit 240 generates service data using sensor data received from the infrastructure sensor 114 via the communication unit 164 (see FIG. 4 ), and provides the service data to the in-vehicle device 100, etc. When the server 106 receives upload data, the server 106 also generates service data using the received upload data.

[0082] The surrounding condition detection unit 242 detects the surrounding conditions of the vehicle (i.e., the vehicle 102) corresponding to the received transmission target data from the sensor data received from the infrastructure sensor 114 and outputs the detection result to the evaluation unit 244. The detection result is stored in the memory 162 (see FIG. 4 ). Data representing the vehicle's driving status (e.g., position, speed, etc.) is repeatedly (e.g., periodically) transmitted from each vehicle to the server 106 and stored in the memory 162. Note that if the source address (e.g., IP address) is subject to change, the data may be transmitted with information (e.g., an ID) identifying the vehicle (specifically, the on-board device). The surrounding condition detection unit 242 can associate the vehicle's location with the uploaded transmission target data using the data's source address and timestamp, and can detect the vehicle's surrounding conditions from the sensor data by referring to the vehicle's location. As described above, the surrounding condition data is data representing the presence of traffic accidents, traffic congestion, blind spots (e.g., intersections with poor visibility), etc.

[0083] The evaluation unit 244 generates an evaluation index in the same manner as described above for the evaluation unit 202. Unlike the evaluation unit 202, the evaluation unit 244 generates an evaluation index that evaluates the output data (i.e., service data) of the service execution unit 240. The output data (i.e., driving assistance information, etc.) of the service execution unit 240 is considered to depend on the surrounding conditions of the vehicle 102 detected by the surrounding condition detection unit 242 and the uploaded transmission target data. For example, if the surrounding condition detection unit 242 detects a traffic accident around the vehicle 102 from sensor data acquired from the infrastructure sensor 114, the service execution unit 240 can generate more effective driving assistance information using the transmission target data (e.g., video data, etc.) uploaded from the vehicle 102. If the output data of the service execution unit 240 when the upload data 230 is not used is Y1 and the output data when the upload data 230 is used is Y2, then Y1 = f(X1) and Y2 = f(X1, X2) can be expressed using a predetermined function (specifically, an algorithm) f. X1 represents a set of driving state data and surrounding condition data, and X2 represents upload data 230. The surrounding condition detection unit 242 generates evaluation indices from Y1 and Y2. If the evaluation index when upload data 230 is not used is Z1 and the evaluation index when upload data 230 is used is Z2, then Z1 = g(Y1) and Z2 = g(Y2) can be expressed using a predetermined function (specifically, a model) g. The evaluation indices are, for example, the detection rate of traffic accidents, etc. (i.e., the degree of missed detection).

[0084] Under the control of the control unit 140, the determination unit 204 of the in-vehicle / out-of-vehicle communication unit 122 reads the evaluation indexes Z1 and Z2 from the evaluation result data 232 and determines whether there is a significant difference between Z1 and Z2. Specifically, if the function g is set so that the more effectively the upload data 230 is used, the larger the evaluation index becomes, the determination unit 204 calculates the difference ΔZ between Z1 and Z2 (ΔZ = Z2 - Z1) and determines whether the difference ΔZ is greater than a predetermined threshold value Th. If the function g is set so that the more effectively the upload data 230 is used, the determination unit 204 calculates the difference ΔZ by ΔZ = Z1 - Z2. In either case, the difference ΔZ is the difference from the evaluation index Z1 when the upload data 230 is not used (i.e., the difference (≧0) based on the evaluation index Z1). If ΔZ > Th, data corresponding to Z2 is read from the driving state data 214 and the surrounding situation data 216 and stored in the memory 142 as the learning result data 220. The data corresponding to Z2 can be identified from the driving condition data 214 and the surrounding condition data 216 using the time stamp.

[0085] As will be described later, the server 106 receives data uploaded from multiple in-vehicle devices, repeatedly (for example, periodically) generates evaluation indices, and transmits them by broadcast. Therefore, assuming that the clocks of the server 106 and the in-vehicle / out-of-vehicle cooperation unit 122 are adjusted to show the same time, among the evaluation indices received by the in-vehicle / out-of-vehicle cooperation unit 122 from the server 106, an evaluation index with a timestamp that is after and close to the time of upload from the in-vehicle / out-of-vehicle cooperation unit 122 can be identified as evaluation index Z2. Other evaluation indices can be designated as evaluation index Z1.

[0086] As described above, the evaluation result data 232 is an evaluation index received by the vehicle interior / exterior linking unit 122 from the server 106, and takes into consideration what is detected by the surrounding condition detection unit 242 of the server 106. In contrast, the surrounding condition data 216 is what is detected by the surrounding condition detection unit 200. The detection result of the surrounding condition detection unit 200 is not necessarily the same as the detection result by the surrounding condition detection unit 242, but since both the surrounding condition detection unit 200 and the surrounding condition detection unit 242 are the surrounding conditions of the vehicle 102, they can be expected to be similar. Therefore, the surrounding condition data 216, which is the detection result of the surrounding condition detection unit 200, can be used as data corresponding to the evaluation index Z2.

[0087] If the evaluation index when the upload data 230 is used is more than Th away from the evaluation index when the upload data 230 is not used, it can be said that the upload data 230 has been used effectively by the server 106 (specifically, the service execution unit 240). Therefore, if a similar state is detected thereafter, the data to be transmitted can be uploaded to the server 106. On the other hand, if the evaluation index when the upload data 230 is used is Th or less away from the evaluation index when the upload data 230 is not used, it cannot be said that the upload data 230 has been used effectively by the server 106. In other words, it can be said that uploading performed in such a state is wasteful.

[0088] During a predetermined learning period, the control unit 140 controls the communication unit 120 to repeatedly (e.g., periodically) upload the data to be transmitted. The surrounding condition detection unit 200 and the determination unit 204 each execute processing at a predetermined timing. Furthermore, the evaluation index transmitted from the server 106 is received by the communication unit 120 and stored in the memory 142 as evaluation result data 232. This accumulates the learning result data 220. After the learning period has elapsed, the control unit 140 reads the latest driving condition data 214 and surrounding condition data 216 and determines whether the read data satisfies the transmission conditions, i.e., whether the read data corresponds to the learning result data 220. If the transmission conditions are satisfied, the control unit 140 controls the communication unit 120 to upload the data to be transmitted to the server 106. If the transmission conditions are not satisfied, the control unit 140 does not perform the upload. Therefore, the in-vehicle / out-of-vehicle linking unit 122 uploads data to be transmitted (i.e., data that can be used for the connected service provided by the server 106, such as sensor data) when there is a high possibility that the data will be used effectively, and does not upload data in other cases. That is, the frequency with which the in-vehicle device 100 uploads data to the server 106 (i.e., roadside device, etc.) can be appropriately reduced without degrading the quality of the connected service, and unnecessary uploads can be suppressed.

[0089] As described above, the communication unit 120 transmits driving condition data representing the driving condition of the vehicle 102 to the server 106, and the evaluation index is generated by the server 106 (specifically, the evaluation unit 244) taking into account the driving condition and the vehicle's surrounding conditions. The learning unit 146 includes a surrounding condition detection unit 200 that generates surrounding condition data representing the surrounding conditions, and a determination unit 204 that determines whether the server 106 effectively used the data to be received by comparing the evaluation index Z2 when the data to be transmitted was transmitted with the evaluation index Z1 when the data to be transmitted was not transmitted during the learning period. Upon determining that the data to be transmitted was effectively used, the determination unit 204 specifies the driving condition data and surrounding condition data when the data to be transmitted was transmitted as transmission conditions. This allows the server 106 to appropriately generate the evaluation index, and the in-vehicle device 100 to appropriately determine transmission conditions for uploading data to the server 106.

[0090] [Operation of the in-vehicle and out-of-vehicle linkage unit (upload)] 10 and 11, the operation of the in-vehicle / out-vehicle linking unit 122 regarding uploading will be described with reference to the functions shown in FIG. 9. The process shown in FIG. 10 is implemented by the control unit 140 reading and executing a predetermined program from the memory 142. The process shown in FIG. 10 starts, for example, when the in-vehicle device 100 is powered on. The learning period T, the upload cycle ΔT1 during the learning period, the learning cycle ΔT2 (e.g., ΔT2>ΔT1), the threshold value Th, and the like are assumed to be stored in advance in the memory 142. In parallel with this program, the control unit 140 executes a program that, upon receiving an evaluation index transmitted from the server 106, adds a timestamp and stores the evaluation result data 232. The sensor data 210 and the upload data 230 are periodically timestamped and stored in the memory 142. The surrounding condition detection unit 200 periodically executes the above-described process and stores the surrounding condition data 216 with a timestamp.

[0091] In step 400, the control unit 140 determines whether or not to perform upload. Specifically, the control unit 140 determines whether or not a period ΔT1 has elapsed since the previous upload was performed. If it is determined that the period ΔT1 has elapsed, the control proceeds to step 402. If not, the control proceeds to step 404. Note that when step 400 is performed for the first time, it is determined that upload is to be performed. The control unit 140 may obtain the current time from, for example, a timer installed in the in-vehicle device 100.

[0092] In step 402, the control unit 140 controls the communication unit 120 to upload the data to be transmitted (i.e., the upload data 230) to the server 106. Thereafter, the control proceeds to step 404.

[0093] In step 404, the control unit 140 determines whether or not to perform learning. Specifically, the control unit 140 determines whether or not a period ΔT2 has elapsed since the previous learning was performed. If it is determined that the period ΔT2 has elapsed, the control proceeds to step 406. If not, the control proceeds to step 418. Note that when step 404 is executed for the first time, it is determined that learning is to be performed.

[0094] In step 406, the control unit 140 learns the upload conditions (i.e., the transmission conditions). Specifically, the control unit 140 executes the process shown in FIG.

[0095] Referring to FIG. 11, in step 430, the control unit 140 reads the evaluation result data 232 stored in the memory 142 and calculates the difference ΔZ (≧0) between the evaluation index Z2 when the data to be transmitted is uploaded and the evaluation index Z1 when the data is not uploaded, as described above. The evaluation index Z2 that is the processing target is the evaluation index Z2 that has not been processed up to that point. The evaluation indexes Z2 and Z1 used to calculate the difference are, for example, the evaluation indexes Z2 and Z1 that are assigned the closest times to each other. If the number of evaluation indexes Z2 is smaller than the number of evaluation indexes Z1, the representative value (e.g., average value) of the evaluation index Z1 that is located between two adjacent evaluation indexes Z2 when arranged in chronological order and one of the two adjacent evaluation indexes Z2 can be used to calculate the difference ΔZ. The control unit 140 stores the calculated difference ΔZ in the memory 142 with time information attached. For example, the time information attached to the evaluation index Z2 used to calculate the difference can be used. Thereafter, control proceeds to step 432.

[0096] In step 432, the control unit 140 determines whether the difference ΔZ calculated in step 430 is greater than the threshold value Th. If ΔZ>Th, the control proceeds to step 434. If not, the control proceeds to step 436.

[0097] In step 434, the control unit 140 identifies the driving condition data and surrounding condition data corresponding to the evaluation index Z2 determined in step 432 to be ΔZ>Th (i.e., the evaluation index at the time of uploading), and stores them as learning result data 220. To identify the driving condition data and surrounding condition data, it is sufficient to compare the time information attached to them with the time information attached to the corresponding evaluation index Z2. The driving condition data and surrounding condition data attached with time information that is the same time as the time information attached to the evaluation index Z2 or that is close within a predetermined range are identified.

[0098] In step 436, the control unit 140 determines whether step 432 has been executed for all of the differences ΔZ calculated in the previously executed step 430. If it is determined that all of them have been executed, the control proceeds to step 408 in Fig. 10. If not, the control returns to step 432.

[0099] Returning to FIG. 10 , in step 408, the control unit 140 determines whether learning is complete. Specifically, the control unit 140 determines whether the learning period T has elapsed. If it is determined that learning is complete, control proceeds to step 410. If not, control proceeds to step 418.

[0100] In step 410, the control unit 140 reads the latest driving condition data 214 and surrounding condition data 216 from the memory 142 and determines whether the upload conditions (i.e., transmission conditions) are met. Specifically, the control unit 140 determines whether a set of values ​​identical to the set of driving condition data 214 and surrounding condition data 216 that have been read, or a set of values ​​close to the set within a predetermined range, is stored in the learning result data 220. If it is determined that the values ​​are stored, control proceeds to step 412. Otherwise, control proceeds to step 414.

[0101] In step 412, the control unit 140 executes the upload in the same manner as in step 402. Thereafter, the control proceeds to step 416.

[0102] If the determination result in step 410 is NO, in step 414, the control unit 140 determines whether or not an end instruction has been received. If it is determined that an end instruction has been received, the program ends. If not, the control returns to step 410. The end instruction is issued, for example, by turning off the in-vehicle device 100.

[0103] In step 416, the control unit 140 determines whether the current learning result is valid. Specifically, the control unit 140 reads from the memory 142 the evaluation index Z2 (i.e., the latest evaluation result data 232) resulting from the use by the server 106 of the data to be transmitted uploaded in the immediately preceding step 412, calculates the difference ΔZ in the same manner as above, and compares it with the threshold value Th. The evaluation index Z1 used as the reference for calculating the difference may be a representative value (e.g., the average, median, etc.) of the evaluation index Z1 during the learning period. If ΔZ>Th, the current learning result (i.e., the learning result data 220 stored in the memory 142) is determined to be valid, and control returns to step 410. Otherwise, the control unit 140 deletes the learning result data 220 stored in the memory 142, and control proceeds to step 418.

[0104] In step 418, the control unit 140 determines whether an end instruction has been received, similar to step 414. If it is determined that an end instruction has been received, the program ends. If not, control returns to step 400.

[0105] [Server Operation (Upload)] With reference to Fig. 12, the operation of server 106 regarding uploading will be described with reference to the functions shown in Fig. 9. The process shown in Fig. 12 is realized by control unit 160 shown in Fig. 4 reading a predetermined program from memory 162 and executing it. The process shown in Fig. 12 is started, for example, by the execution of an application for providing a service (specifically, corresponding to service execution unit 240). In parallel with this program, control unit 160 executes the application for providing the service, a program for analyzing sensor data received from infrastructure sensors 114 and the like to detect the surrounding conditions of the vehicle (corresponding to surrounding condition detection unit 242), and the like.

[0106] In step 500, the control unit 160 determines whether or not data uploaded from the in-vehicle device has been received. If it is determined that data has been received, control proceeds to step 502. If not, control proceeds to step 504.

[0107] In step 502, the control unit 160 passes the data received in step 500 to the application. Then, control proceeds to step 504. The application uses the received data to generate service data. Note that the server 106 receives data uploaded from multiple in-vehicle devices.

[0108] In step 504, the control unit 160 determines whether or not to calculate an evaluation index. For example, the evaluation index is calculated periodically. For example, the control unit 160 determines whether or not a predetermined period (for example, period ΔT3) has elapsed since the previous calculation of the evaluation index. If it is determined that the predetermined period has elapsed, control proceeds to step 506. Otherwise, control proceeds to step 510. Note that when step 504 is executed for the first time, it is determined that the evaluation index is to be calculated.

[0109] In step 506, the control unit 160 calculates the evaluation index, which corresponds to the function of the evaluation unit 244 described above. Thereafter, the control proceeds to step 508.

[0110] In step 508, the control unit 160 assigns a timestamp to the evaluation index calculated in step 506 and transmits the evaluation index by broadcast via the communication unit 164 (see FIG. 4). As described above, the in-vehicle / out-of-vehicle cooperation unit 122 receives the transmitted evaluation index and stores it in the memory 142 as evaluation result data 232. Thereafter, the control proceeds to step 510.

[0111] In step 510, the control unit 160 determines whether an instruction to terminate has been received. If it is determined that an instruction to terminate has been received, the program terminates. If not, control returns to step 500. An instruction to terminate is issued, for example, by the administrator operating the operation unit (e.g., keyboard, mouse, etc.) of the server 106 to terminate the application.

[0112] By repeating the processes of steps 400 to 408 in FIG. 10 and steps 500 to 508 in FIG. 12, uploading and learning of upload conditions (transmission conditions) are periodically performed until learning is completed (the learning period has elapsed). While the above description describes uploading and learning being periodically performed, this is not limiting. It is sufficient that the frequency of repeated uploading is greater than the frequency of repeated learning. After learning is completed, by repeating the processes of steps 410 to 416 in FIG. 10 and steps 500 to 508 in FIG. 12, it is determined whether to perform uploading using the learning results, and uploading is performed only when the upload conditions are satisfied. Therefore, the frequency at which the on-board device 100 uploads data to the server 106 (i.e., a roadside device, etc.) can be appropriately reduced without degrading the quality of connected services, thereby preventing unnecessary uploads.

[0113] Furthermore, if the learning result data 220 (see FIG. 9) stored in memory 142 becomes invalid (i.e., if there is no longer a significant difference between evaluation index Z2 and evaluation index Z1 (specifically, the judgment result in step 416 is NO)), re-learning is executed. That is, control returns to step 400, and uploading and learning of upload conditions (i.e., transmission conditions) are periodically executed again. As a result, if the learning result becomes invalid, re-learning is executed promptly, and appropriate transmission conditions can be determined again.

[0114] As described above, the server 106 includes the service execution unit 240 that executes a predetermined service, the evaluation unit 244 that generates an evaluation index that indicates the degree to which the data to be transmitted has been used by the service execution unit 240, and the communication unit 164 (see FIG. 4) that transmits the evaluation index to the in-vehicle device 100. This allows the determination unit 204 to appropriately determine the transmission conditions for uploading data to the server 106, and the in-vehicle device 100 to appropriately reduce the frequency of uploading data to the server 106, thereby suppressing unnecessary uploads.

[0115] In the above, the same symbols are used for the learning period for uploading, the upload cycle during the learning period, the learning cycle, and the threshold value for downloading, but the actual numerical values ​​are arbitrary. The same or different values ​​may be used for downloading and uploading. Furthermore, learning may be completed not only when the learning period has elapsed, but also when a predetermined amount of learning result data has been collected.

[0116] The server may manage uploaded data to be transmitted in association with its source (i.e., the in-vehicle device). In this case, if the evaluation index is broadcast with information identifying the source, each in-vehicle device can identify whether the received evaluation index is evaluation index Z2 that takes into account the data it uploaded, or another evaluation index Z1.

[0117] When vehicles and on-board devices have the same system configuration and are used in the same traffic environment (e.g., urban areas, depopulated areas, etc.), the learning result data will be the same or similar. Therefore, the learning result data may be shared. For example, the learning results (i.e., learning result data) of one vehicle may be stored in the memory of the on-board device of another vehicle, and the appropriateness of downloading or uploading may be determined in the other vehicle. If the learning results are not appropriate, the on-board device of the other vehicle may simply perform re-learning.

[0118] In the above description, the ECU that uses the downloaded data (i.e., the data to be received) is an autonomous driving ECU, and the evaluation index depends on the vehicle's driving state and the vehicle's surrounding conditions. However, this is not limiting. An appropriate evaluation index may be adopted depending on the function of the ECU that uses the downloaded data. The download conditions (i.e., reception conditions) can be determined by detecting the information taken into account when generating the evaluation index using a sensor 124 or the like, and performing learning using the difference in the evaluation index depending on whether or not the downloaded data is used, as described above. The same applies to learning the upload conditions.

[0119] [Variations] In the above description, the in-vehicle device (specifically, the in-vehicle / out-of-vehicle linking unit 122) learns the upload conditions, but this is not limiting. The server may learn the upload conditions of the in-vehicle device and transmit the learning results to the in-vehicle device. The in-vehicle / out-of-vehicle linking unit of the in-vehicle device does not perform learning regarding uploading, but instead uses the learning results received from the server to determine whether uploading is appropriate.

[0120] With reference to FIG. 13, the functions of the in-vehicle-external linking unit 122 and the server 106 according to the modified example will be described. The in-vehicle-external linking unit 122 is configured as shown in FIG. 6. In FIG. 13, the same configuration as in FIG. 6 is shown for the in-vehicle-external linking unit 122, and the configuration not used during uploading according to the modified example (i.e., the evaluation unit 202, the determination unit 204, the download data 212, and the evaluation result data 218) is indicated by dashed lines. As described above, an evaluation index indicating the degree to which the data to be transmitted (i.e., the upload data) is effectively used is used for the evaluation index. As will be described later, the server 106 calculates the evaluation index for uploading and performs learning, and therefore the server 106 in FIG. 13 shows the configuration required for this. Note that the "learning result data" stored in the memory 142 is different from the learning result data 220 shown in FIG. 6 and is received from the server 106, and is therefore indicated by a different reference numeral "234" from that shown in FIG. 6. In the following, the same functions as those of the in-vehicle / out-of-vehicle cooperation unit 122 and the server 106 when the in-vehicle / out-of-vehicle cooperation unit 122 learns will not be described repeatedly, and differences will mainly be described.

[0121] In addition to the data shown in Fig. 6, memory 142 also stores upload data 230, similar to Fig. 9. Upload data 230 is data to be uploaded and transmitted, and is, for example, data (e.g., video data) from sensor data 210 (i.e., output data of sensor 124) that can be effectively used by server 106.

[0122] The server 106 includes a service execution unit 240, a surrounding condition detection unit 242, an evaluation unit 244, and a determination unit 246 to handle uploads from the vehicle interior / exterior cooperation unit 122. Of these, the service execution unit 240, the surrounding condition detection unit 242, and the evaluation unit 244 function in the same manner as those shown in FIG. 9 . That is, the service execution unit 240 executes connected services provided by the server 106 (for example, providing driving assistance information, remote monitoring, remote control, etc.). The surrounding condition detection unit 242 detects the surrounding conditions of the vehicle (i.e., the vehicle 102) corresponding to the uploaded data to be transmitted from the sensor data received from the infrastructure sensor 114, and outputs the detection result to the evaluation unit 244. The evaluation unit 244 generates an evaluation index that evaluates the output data (i.e., service data) of the service execution unit 240.

[0123] The determination unit 246 acquires the evaluation indexes Z1 and Z2 from the evaluation unit 244, and determines whether there is a significant difference between Z1 and Z2, similar to the determination unit 204 of the vehicle interior / exterior linkage unit 122 shown in FIG. 6. The determination unit 246 calculates the difference ΔZ between Z1 and Z2, and determines whether the difference ΔZ is greater than a predetermined threshold value Th. If ΔZ>Th, the determination unit 246 reads the vehicle running state data and surrounding situation data corresponding to Z2 from the memory 162 (see FIG. 4) and transmits them to the vehicle 102 (specifically, the vehicle interior / exterior linkage unit 122). A timestamp can be used to identify the data corresponding to Z2 from the running state data and surrounding situation data stored in the memory 162.

[0124] For example, as shown in FIG. 10 , the in-vehicle / out-of-vehicle linking unit 122 repeatedly (e.g., periodically) uploads data. The server 106 repeatedly (e.g., periodically) calculates the evaluation index as described above for a predetermined period (learning period), performs learning using the evaluation index, and transmits the learning result to the in-vehicle / out-of-vehicle linking unit 122. As a result, learning result data 234 is accumulated. After the learning period has elapsed, the in-vehicle / out-of-vehicle linking unit 122 determines whether uploading is appropriate using the learning result data 234 as a transmission condition. That is, the control unit 140 of the in-vehicle / out-of-vehicle linking unit 122 reads the latest driving state data 214 and surrounding situation data 216 from the memory 142 and determines whether the learning result data 234 (i.e., the transmission condition) is satisfied. Specifically, the control unit 140 determines whether the learning result data 234 contains the same values ​​as the set of driving state data 214 and surrounding situation data 216 read out, or a set of values ​​close within a predetermined range. If it is determined that the values ​​are stored, uploading is performed. Otherwise, uploading is not performed. In this way, the in-vehicle device 100 (specifically, the in-vehicle / out-vehicle linking unit 122) that receives the transmission conditions can appropriately reduce the frequency at which the in-vehicle device 100 uploads data to the server 106 (i.e., a roadside device, etc.), thereby suppressing unnecessary uploads.

[0125] In the modified example, it is also preferable to perform re-learning if the learning result is no longer appropriate. To do so, similar to the configuration shown in FIG. 9 , the evaluation unit 244 transmits an evaluation index to the in-vehicle device 100 and stores it in the memory 142 (i.e., the evaluation result data 218). For example, the control unit 140 of the in-vehicle / external cooperation unit 122 determines whether the upload is appropriate using the learning result data 234 (i.e., the learning result by the server 106), and then determines whether the current learning result is valid by referring to the evaluation result data 218, similar to step 416 shown in FIG. 10 . If the current learning result is determined to be invalid, the in-vehicle / external cooperation unit 122 deletes the current learning result data 234, transmits a re-learning request to the server 106, and then repeatedly (e.g., periodically) performs uploading for a predetermined period (i.e., the learning period). As a result, the server 106 (specifically, the determination unit 246) that received the re-learning request again performs re-learning as described above and transmits the learning result to the in-vehicle / external cooperation unit 122. As a result, if the learning result becomes invalid, the in-vehicle / out-of-vehicle cooperation unit 122 can promptly cause the server 106 to re-learn and determine appropriate transmission conditions again.

[0126] In the above, the reception conditions and transmission conditions are individually collected (and learned) to generate the learning result data shown in FIGS. 6, 9, and 13. However, this is not limiting. Machine learning may be used to determine whether a situation is suitable for reception and transmission (hereinafter, referred to as the appropriateness of reception and transmission). For example, learning may be performed using a neural network. In this case, learning data and a model (i.e., a neural network) are prepared. For example, in FIG. 6, the model that determines the appropriateness of reception (i.e., downloading) is included in the in-vehicle / external link unit 122 as a program that replaces the learning result data 220. The vehicle's driving status (e.g., position, speed, etc.) and the vehicle's surrounding conditions (e.g., traffic jams, accidents, blind spots, etc.) are used as input data to the model's input layer, and the output data of the model's output layer indicates whether or not a download has been performed. The model is trained using this learning data. The presence or absence of a download can be quantified as "1" if a download has been performed and "0" if not. The number of neurons in the output layer may be one. The input speed can be quantified using the speed value itself (or its normalized value (i.e., a value obtained by multiplying the speed itself by a constant to make the magnitude of each input value uniform)). Regarding the quantification of the vehicle's surrounding environment, for example, the number of vehicles around the vehicle (e.g., within a specified area) can be used for congestion, and for accidents, a value of "1" can be used if an accident has occurred around the vehicle (e.g., within a specified area), and a value of "0" can be used if no accident has occurred. Regarding blind spots, the angle, the size (i.e., area) of the blind spot area, etc. (or their normalized values) can be used for quantification. These values ​​can be obtained, for example, by analyzing video data obtained by a sensor on the vehicle. In learning, a sigmoid function or the like can be used as the activation function of the synapses in the intermediate layer, and the activation function of the synapses in the intermediate layer can be adjusted using the above-mentioned training data (corresponding to the training data) to minimize the loss function. Similarly, in the case of uploading, learning can be performed by using the output of the output layer as the presence or absence of upload. The presence or absence of upload can be quantified as "1" if uploading occurs, and "0" if not.

[0127] As described above, the learning unit 146 repeatedly (e.g., periodically) downloads for a predetermined period of time, compares the difference ΔZ in the evaluation index with the threshold value Th, and determines whether or not the download is appropriate (“1” or “0”). The results, along with the corresponding vehicle driving conditions (e.g., location and speed) and vehicle surrounding conditions (e.g., traffic jams, accidents, blind spots, etc.), may be collected as learning data. For example, the control unit 140 uses the collected learning data to train a model as described above. When the vehicle driving conditions and vehicle surrounding conditions are input, the trained model outputs data indicating whether or not a download is appropriate. Therefore, the control unit 140 reads the latest vehicle driving conditions and vehicle surrounding conditions from memory 142 and inputs them into the trained model. If the output value of the trained model is “1,” the download is performed; if the output value is “0,” the download is not performed. This allows the appropriateness of a download to be automatically determined without collecting and storing download conditions (i.e., reception conditions) as shown in FIGS. 6 and 7. Therefore, the in-vehicle / out-of-vehicle linking unit 122 downloads the data to be received (i.e., data for the connected service provided by the server 106, such as driving assistance information) when there is a high possibility that the data will be used effectively, and does not download the data in other cases. That is, the frequency with which the in-vehicle device 100 downloads data for the service provided by the server 106 (i.e., a roadside device, etc.) can be appropriately reduced without degrading the quality of the connected service, and unnecessary downloads can be suppressed.

[0128] Similarly, with regard to uploading, uploading is performed repeatedly (e.g., periodically) for a predetermined period, and the difference ΔZ in the evaluation index is compared with a threshold value Th to determine whether uploading is appropriate (“1” or “0”). The result, along with the corresponding vehicle driving state and vehicle surrounding conditions, is collected as learning data. A model is trained using the collected learning data. A model that determines whether transmission (i.e., uploading) is appropriate is included in the in-vehicle / external linking unit 122 as a program substituting the learning result data 220 in FIG. 9, for example. Upon inputting the vehicle driving state and vehicle surrounding conditions, the trained model outputs data indicating whether uploading is appropriate (i.e., appropriate). Therefore, the control unit 140 reads the latest vehicle driving state and vehicle surrounding conditions from memory 142 and inputs them into the trained model. If the output value of the trained model is “1,” uploading is performed; if the output value is “0,” uploading is not performed. This allows automatic determination of whether uploading is appropriate without collecting and storing upload conditions (i.e., transmission conditions) as shown in FIGS. 9 and 10. Therefore, the in-vehicle / out-of-vehicle linking unit 122 uploads data to be transmitted (i.e., data that can be used for the connected service provided by the server 106, such as sensor data) when there is a high possibility that the data will be used effectively, and does not upload data in other cases. That is, the frequency with which the in-vehicle device 100 uploads data to the server 106 (i.e., roadside device, etc.) can be appropriately reduced without degrading the quality of the connected service, and unnecessary uploads can be suppressed.

[0129] Furthermore, learning regarding downloading and uploading may be performed simultaneously. In this case, the model output must represent three types: download, upload, and do neither, so three neurons are used in the model's output layer. A softmax function can be used as the activation function for the synapses in the output layer, and a sigmoid function or similar can be used as the activation function for the synapses in the intermediate layer. By using the training data to adjust the activation functions for the synapses in the intermediate layer so that the loss function is minimized, the output from each synapse in the output layer will represent the corresponding three types of probability, and selecting the one corresponding to the maximum value will determine one of the three types (i.e., download, upload, or do neither).

[0130] In addition, in the modified example (see FIG. 13 ), a model (specifically, a program) is prepared in the server 106, and after machine learning is performed as described above, the in-vehicle / out-of-vehicle cooperation unit 122 receives the trained model from the server 106 and uses it instead of the training result data 234. Alternatively, a model having the same configuration as the model of the server 106 may be prepared in advance in the in-vehicle / out-of-vehicle cooperation unit 122, and parameters of the trained model of the server 106 (i.e., parameters specifying the activation function) may be received from the server 106 and applied to the model of the in-vehicle / out-of-vehicle cooperation unit 122.

[0131] Reinforcement learning may be performed for a predetermined period of time without using training data. Furthermore, machine learning, such as a support vector machine, may be used instead of a neural network. When determining download conditions or upload conditions, the process comes down to classifying a set of feature quantities (e.g., the vehicle's driving state and the vehicle's surrounding conditions) (i.e., multidimensional data) into two types (i.e., download (or upload) or not) using a support vector machine. When determining one of three types (download, upload, or neither), the process comes down to classifying a set of feature quantities (e.g., the vehicle's driving state and the vehicle's surrounding conditions) (i.e., multidimensional data) into three types using a support vector machine.

[0132] Each process (each function) in the above-described embodiments may be implemented by a processing circuit including one or more processors. The processing circuit may be configured by an integrated circuit or the like that combines one or more memories, various analog circuits, and various digital circuits in addition to the one or more processors. The one or more memories store programs (instructions) that cause the one or more processors to execute the respective processes. The one or more processors may execute the respective processes according to the programs read from the one or more memories, or according to logic circuits pre-designed to execute the respective processes. The processor may be any of various processors suitable for computer control, such as a CPU, a GPU (Graphics Processing Unit), a DSP (Digital Signal Processor), an FPGA (Field Programmable Gate Array), or an ASIC (Application Specific Integrated Circuit).

[0133] In addition, a recording medium can be provided that stores a program that causes a computer to execute the processing of the in-vehicle device 100 (specifically, the processing executed by the in-vehicle / out-vehicle linking unit 122 (for example, the processing shown in Figures 7, 8, 10, and 11). The recording medium is, for example, an optical disc (such as a DVD (Digital Versatile Disc)) or a removable semiconductor memory (such as a USB (Universal Serial Bus) memory). Although a computer program can be transmitted over a communication line, the recording medium refers to a non-transitory recording medium. By having the computer mounted on the vehicle load the program stored in the recording medium, the computer can transmit data that can be effectively used by a service provided by the external device, taking into account delay time and communication bandwidth when the in-vehicle device uploads data to an external device such as a roadside device, as described above.

[0134] (Appendix 1) That is, the computer-readable non-transitory recording medium is The computer installed in the vehicle a communication function for receiving data to be received from a roadside device located outside the vehicle; a learning function that learns reception conditions for the data to be received by evaluating the extent to which the data to be received has been used by a function control device mounted on the vehicle, and the learning function includes a function of identifying the reception conditions during a learning period in which the reception target data is repeatedly received by the communication function, The communication function stores a computer program including a function of receiving the data to be received depending on whether the reception conditions are satisfied after the reception conditions are identified by the learning function.

[0135] (Appendix 2) In addition, the computer-readable non-transitory recording medium is The computer installed in the vehicle a communication function for receiving data to be received from a roadside device located outside the vehicle; a learning function for learning whether or not the reception of the target data by the communication function is appropriate; The learning function is an acquisition function for acquiring output data of a function control device mounted on the vehicle and running state data representing a running state of the vehicle; a surrounding situation detection function for generating surrounding situation data representing the surrounding situation of the vehicle; an evaluation function that generates an evaluation index for evaluating the running of the vehicle from the output data during a predetermined period in which the reception target data is repeatedly received by the communication function; a determination function that determines whether the data to be received has been effectively used by the function control device by comparing the evaluation index when the data to be received is received with the evaluation index when the data to be received is not received during the predetermined period; a function of performing machine learning using learning data to generate a model that outputs data indicating the suitability of reception in accordance with input data including the driving state data and the surrounding situation data; the communication function includes a function of receiving the data to be received in response to output data of the model after learning; The learning data includes the driving condition data and the surrounding situation data collected during the specified period as the input data, and the judgment results by the judgment function collected during the specified period as the output data of the model. A computer program is stored.

[0136] (Appendix 3) In addition, the computer-readable non-transitory recording medium is The computer installed in the vehicle a communication function for transmitting data to be transmitted to a roadside device that is a device located outside the vehicle, and receiving, from the roadside device, an evaluation index that indicates the degree to which the data to be transmitted has been used by the roadside device; a learning function for learning a transmission condition of the transmission target data using the evaluation index; the learning function includes a function of identifying the transmission condition during a learning period in which the transmission target data is repeatedly transmitted by the communication function; The communication function stores a computer program including a function of transmitting the data to be transmitted depending on whether the transmission condition is satisfied after the transmission condition is identified by the learning function.

[0137] (Appendix 4) In addition, the computer-readable non-transitory recording medium is The computer installed in the vehicle a communication function for transmitting data to be transmitted to a roadside device that is a device located outside the vehicle, and receiving, from the roadside device, an evaluation index that evaluates a service provided by the roadside device; a learning function for learning whether or not the transmission of the transmission target data by the communication function is appropriate; the communication function includes a function of transmitting driving state data representing a driving state of the vehicle to the roadside device; the evaluation index is generated by the roadside device taking into consideration the driving state data and a surrounding situation of the vehicle; The learning function is a surrounding situation detection function for generating surrounding situation data representing the surrounding situation; a determination function that determines whether the data to be transmitted has been effectively used by the roadside device by comparing the evaluation index when the data to be transmitted is transmitted with the evaluation index when the data to be transmitted is not transmitted during a predetermined period in which the data to be transmitted is repeatedly transmitted by the communication function; and a function of performing machine learning using learning data to generate a model that outputs data indicating whether the transmission is appropriate or not in response to input data including the driving state data and the surrounding situation data; The communication function further includes a function of transmitting the transmission target data in response to output data of the model after learning, The learning data includes the driving condition data and the surrounding situation data collected during the specified period as the input data, and the judgment results by the judgment function collected during the specified period as the output data of the model. A computer program is stored.

[0138] Although the present disclosure has been described above by explaining the embodiments, the above-described embodiments are merely examples, and the present disclosure is not limited to only the above-described embodiments. The scope of the present disclosure is defined by the claims in the scope of the claims, taking into consideration the description of the detailed description of the invention, and includes all modifications within the meaning and scope equivalent to the wordings described therein. [Explanation of symbols]

[0139] 100, 110 In-vehicle equipment 102, 112 cars 104 Base station 106 Server 108 Network 114 Infrastructure Sensors 120, 164 Communications Department 122 In-vehicle and out-vehicle coordination department 124 sensors 126 Autonomous Driving ECU 128 Drive ECU Buses 130 and 166 140, 160 control unit 142, 162 memory 144 I / F section 146 Learning Department 200, 242 Surrounding situation detection unit 202, 244 Evaluation Department 204, 246 Judgment section 210 Sensor Data 212 Download Data 214 Driving condition data 216 Surrounding situation data 218, 232 Evaluation result data 220, 234 Learning result data 230 Upload Data 240 Service Execution Department 300, 302, 304, 306, 308, 310, 312, 314, 316, 318, 330, 332, 334, 336, 400, 402, 404, 406, 408, 410, 412, 414, 416, 418, 430, 432, 434, 436, 500, 502, 504, 506, 508, 510 steps

Claims

1. An in-vehicle device mounted on a vehicle, a communication unit that receives data to be received from a roadside device that is an external device of the vehicle; a learning unit that learns a reception condition for the reception target data by evaluating the degree to which the reception target data received by the communication unit has been used by a function control device mounted on the vehicle, the learning unit identifies the reception conditions during a learning period in which the communication unit repeatedly receives the reception target data; The communication unit receives the data to be received depending on whether the reception conditions are satisfied after the learning unit identifies the reception conditions.

2. The learning unit an acquisition unit that acquires output data of the function control device and running state data that represents a running state of the vehicle; a surrounding situation detection unit that generates surrounding situation data representing the surrounding situation of the vehicle; an evaluation unit that generates a first evaluation index for evaluating the running of the vehicle from the output data; a determination unit that determines whether the data to be received has been effectively used by the function control device by comparing the first evaluation index when the data to be received is received with the first evaluation index when the data to be received is not received during the learning period, 2. The in-vehicle device according to claim 1, wherein, upon determining that the data to be received has been effectively used, the determination unit specifies, as the reception conditions, the driving state data and the surrounding situation data at the time the data to be received was received.

3. The in-vehicle device according to claim 2 , wherein the first evaluation index includes at least one of comfort, traffic efficiency, and safety related to the travel of the vehicle.

4. the determination unit determines whether the function control device has effectively utilized the data to be received by determining whether a difference between the first evaluation index when the data to be received is not received and the first evaluation index when the data to be received is equal to or greater than a predetermined value greater than 0 during the learning period; 4. The in-vehicle device according to claim 2, wherein after the learning unit has identified the reception conditions, when the difference becomes smaller than the predetermined value, the learning unit causes the communication unit to repeatedly receive the data to be received and executes the process of identifying the reception conditions again.

5. 4. The in-vehicle device according to claim 1, wherein, in response to an update of the function control device, the learning unit repeatedly receives the reception target data and executes the process of identifying the reception conditions again.

6. The communication unit further comprises: Transmitting data to be transmitted to the roadside device; receiving, from the roadside device, a second evaluation index that evaluates a service provided by the roadside device; The learning unit further learning a transmission condition of the data to be transmitted using the second evaluation index; specifying the transmission condition during a period in which the communication unit repeatedly transmits the transmission target data; The in-vehicle device according to claim 1 , wherein the communication unit transmits the data to be transmitted depending on whether the transmission condition is satisfied after the learning unit identifies the transmission condition.

7. An in-vehicle device mounted on a vehicle, a communication unit that receives data to be received from a roadside device that is an external device of the vehicle; a learning unit that learns whether or not reception of the reception target data by the communication unit is appropriate, The learning unit an acquisition unit that acquires output data of a function control device mounted on the vehicle and running state data that represents a running state of the vehicle; a surrounding situation detection unit that generates surrounding situation data representing the surrounding situation of the vehicle; an evaluation unit that generates an evaluation index for evaluating the running of the vehicle from the output data during a predetermined period in which the communication unit repeatedly receives the reception target data; a determination unit that determines whether the data to be received has been effectively used by the function control device by comparing the evaluation index when the data to be received is received with the evaluation index when the data to be received is not received during the predetermined period; a model that outputs data indicating whether the reception is appropriate or not in response to input data including the driving state data and the surrounding situation data; the learning unit performs machine learning on the model using learning data; the communication unit receives the data to be received in accordance with output data of the model after learning; an on-board device, wherein the learning data includes the driving condition data and the surrounding situation data collected during the predetermined period as the input data, and includes a determination result by the determination unit collected during the predetermined period as the output data of the model.

8. An in-vehicle device mounted on a vehicle, a communication unit that transmits transmission target data to a roadside device that is a device located outside the vehicle and receives, from the roadside device, an evaluation index that evaluates a service provided by the roadside device; a learning unit that learns a transmission condition of the transmission target data using the evaluation index; the learning unit identifies the transmission condition during a learning period in which the communication unit repeatedly transmits the transmission target data; The communication unit transmits the data to be transmitted depending on whether the transmission condition is satisfied after the learning unit identifies the transmission condition.

9. the communication unit further transmits driving state data representing a driving state of the vehicle to the roadside device; the evaluation index is generated by the roadside device taking into consideration the driving state data and a surrounding situation of the vehicle; The learning unit a surrounding situation detection unit that generates surrounding situation data representing the surrounding situation; a determination unit that determines whether the data to be transmitted has been effectively used by the roadside device by comparing the evaluation index when the data to be transmitted is transmitted with the evaluation index when the data to be transmitted is not transmitted during the learning period, 9. The in-vehicle device according to claim 8, wherein, upon determining that the data to be transmitted has been effectively used, the determination unit specifies, as the transmission conditions, the driving state data and the surrounding situation data at the time the data to be transmitted was transmitted.

10. the determination unit determines whether or not a difference between the evaluation index when the data to be transmitted is transmitted and the evaluation index when the data to be transmitted is not transmitted during the learning period is equal to or greater than a predetermined value greater than 0, thereby determining whether or not the data to be transmitted is effectively used by the roadside device; 10. The in-vehicle device according to claim 9, wherein after the transmission conditions have been identified by the learning unit, when the difference becomes smaller than the predetermined value, the learning unit causes the communication unit to repeatedly transmit the data to be transmitted, and executes the process of identifying the transmission conditions again.

11. An in-vehicle device mounted on a vehicle, a communication unit that transmits transmission target data to a roadside device that is a device located outside the vehicle and receives, from the roadside device, an evaluation index that evaluates a service provided by the roadside device; a learning unit that learns whether or not transmission of the transmission target data by the communication unit is appropriate, the communication unit further transmits driving state data representing a driving state of the vehicle to the roadside device; the evaluation index is generated by the roadside device taking into consideration the driving state data and a surrounding situation of the vehicle; The learning unit a surrounding situation detection unit that generates surrounding situation data representing the surrounding situation; a determination unit that determines whether the data to be transmitted has been effectively used by the roadside device by comparing the evaluation index when the data to be transmitted is transmitted with the evaluation index when the data to be transmitted is not transmitted during a predetermined period in which the communication unit repeatedly transmits the data to be transmitted; a model that outputs data indicating whether the transmission is appropriate or not in response to input data including the driving state data and the surrounding situation data, the learning unit performs machine learning on the model using learning data; the communication unit transmits the transmission target data in accordance with output data of the model after learning; an on-board device, wherein the learning data includes the driving condition data and the surrounding situation data collected during the predetermined period as the input data, and includes a determination result by the determination unit collected during the predetermined period as the output data of the model.

12. a communication unit that communicates with an in-vehicle device mounted in the vehicle and receives data to be transmitted; a service execution unit that executes a predetermined service; a learning unit that learns a transmission condition for the transmission target data by the in-vehicle device by evaluating a degree to which the transmission target data received by the communication unit has been used by the service execution unit, the learning unit identifies the transmission condition during a learning period in which the communication unit receives the transmission target data that is repeatedly transmitted; The communication unit transmits the transmission conditions identified by the learning unit to the vehicle-mounted device.

13. A method for controlling an in-vehicle device mounted on a vehicle, comprising: a communication step of receiving data to be received from a roadside device that is a device located outside the vehicle; a learning step of learning a reception condition for the reception target data by evaluating a degree to which the reception target data received in the communication step has been used by a function control device mounted on the vehicle, the learning step includes a step of identifying the reception conditions during a learning period in which the reception target data is repeatedly received by the communication step, The control method, wherein the communication step includes a step of receiving the data to be received depending on whether the reception conditions are satisfied after the reception conditions are identified in the learning step.

14. A method for controlling an in-vehicle device mounted on a vehicle, comprising: a communication step of receiving data to be received from a roadside device that is a device located outside the vehicle; a learning step of learning whether or not the reception of the data to be received by the communication step is appropriate, The learning step an acquiring step of acquiring output data of a function control device mounted on the vehicle and running state data representing a running state of the vehicle; a surrounding situation detection step of generating surrounding situation data representing a surrounding situation of the vehicle; an evaluation step of generating an evaluation index for evaluating the running of the vehicle from the output data during a predetermined period in which the reception target data is repeatedly received by the communication step; a determination step of determining whether the data to be received has been effectively used by the function control device by comparing the evaluation index when the data to be received is received with the evaluation index when the data to be received is not received during the predetermined period; a step of performing machine learning using learning data to generate a model that outputs data indicating the suitability of reception in response to input data including the driving state data and the surrounding situation data; the communicating step includes a step of receiving the data to be received in response to output data of the model after training; a control method in which the learning data includes the driving condition data and the surrounding situation data collected during the predetermined period as the input data, and includes the judgment results by the judgment step collected during the predetermined period as the output data of the model.

15. A method for controlling an in-vehicle device mounted on a vehicle, comprising: a communication step of transmitting data to be transmitted to a roadside device that is a device located outside the vehicle, and receiving, from the roadside device, an evaluation index that indicates the degree to which the data to be transmitted has been used by the roadside device; a learning step of learning a transmission condition of the transmission target data using the evaluation index; the learning step includes a step of identifying the transmission condition during a learning period in which the transmission target data is repeatedly transmitted by the communication step; The control method, wherein the communication step includes a step of transmitting the data to be transmitted depending on whether the transmission condition is satisfied after the transmission condition is identified in the learning step.

16. A method for controlling an in-vehicle device mounted on a vehicle, comprising: a communication step of transmitting transmission target data to a roadside device that is a device located outside the vehicle, and receiving, from the roadside device, an evaluation index that evaluates a service provided by the roadside device; a learning step of learning whether or not the transmission of the transmission target data by the communication step is appropriate, the communicating step includes a step of transmitting driving state data representing a driving state of the vehicle to the roadside device; the evaluation index is generated by the roadside device taking into consideration the driving state data and a surrounding situation of the vehicle; The learning step a surrounding situation detection step of generating surrounding situation data representing the surrounding situation; a determination step of determining whether the data to be transmitted has been effectively used by the roadside device by comparing the evaluation index when the data to be transmitted is transmitted with the evaluation index when the data to be transmitted is not transmitted during a predetermined period of time during which the data to be transmitted is repeatedly transmitted by the communication step; a step of performing machine learning using learning data to generate a model that outputs data indicating whether the transmission is appropriate or not in response to input data including the driving state data and the surrounding situation data; the communicating step further includes a step of transmitting the data to be transmitted in response to output data of the model after training; a control method in which the learning data includes the driving condition data and the surrounding situation data collected during the predetermined period as the input data, and includes the judgment results by the judgment step collected during the predetermined period as the output data of the model.

17. The computer installed in the vehicle a communication function for receiving data to be received from a roadside device located outside the vehicle; a learning function that learns reception conditions for the data to be received by evaluating the extent to which the data to be received has been used by a function control device mounted on the vehicle, and the learning function includes a function of identifying the reception conditions during a learning period in which the reception target data is repeatedly received by the communication function, The computer program includes a function in which, after the learning function identifies the reception conditions, the communication function receives the data to be received depending on whether the reception conditions are met.

18. The computer installed in the vehicle a communication function for receiving data to be received from a roadside device located outside the vehicle; a learning function for learning whether or not the reception of the target data by the communication function is appropriate; The learning function is an acquisition function for acquiring output data of a function control device mounted on the vehicle and running state data representing a running state of the vehicle; a surrounding situation detection function for generating surrounding situation data representing the surrounding situation of the vehicle; an evaluation function that generates an evaluation index for evaluating the running of the vehicle from the output data during a predetermined period in which the reception target data is repeatedly received by the communication function; a determination function that determines whether the data to be received has been effectively used by the function control device by comparing the evaluation index when the data to be received is received with the evaluation index when the data to be received is not received during the predetermined period; a function of performing machine learning using learning data to generate a model that outputs data indicating the suitability of reception in accordance with input data including the driving state data and the surrounding situation data; the communication function includes a function of receiving the data to be received in response to output data of the model after learning; The learning data includes the driving condition data and the surrounding situation data collected during the specified period as the input data, and includes the judgment results by the judgment function collected during the specified period as the output data of the model.

19. The computer installed in the vehicle a communication function for transmitting data to be transmitted to a roadside device that is a device located outside the vehicle, and receiving, from the roadside device, an evaluation index that indicates the degree to which the data to be transmitted has been used by the roadside device; a learning function for learning a transmission condition of the transmission target data using the evaluation index; the learning function includes a function of identifying the transmission condition during a learning period in which the transmission target data is repeatedly transmitted by the communication function; The computer program includes a function in which, after the learning function identifies the transmission conditions, the communication function transmits the data to be transmitted depending on whether the transmission conditions are satisfied.

20. The computer installed in the vehicle a communication function for transmitting data to be transmitted to a roadside device that is a device located outside the vehicle, and receiving, from the roadside device, an evaluation index that evaluates a service provided by the roadside device; a learning function for learning whether or not the transmission of the transmission target data by the communication function is appropriate; the communication function includes a function of transmitting driving state data representing a driving state of the vehicle to the roadside device; the evaluation index is generated by the roadside device taking into consideration the driving state data and a surrounding situation of the vehicle; The learning function is a surrounding situation detection function for generating surrounding situation data representing the surrounding situation; a determination function that determines whether the data to be transmitted has been effectively used by the roadside device by comparing the evaluation index when the data to be transmitted is transmitted with the evaluation index when the data to be transmitted is not transmitted during a predetermined period in which the data to be transmitted is repeatedly transmitted by the communication function; and a function of performing machine learning using learning data to generate a model that outputs data indicating whether the transmission is appropriate or not in response to input data including the driving state data and the surrounding situation data; The communication function further includes a function of transmitting the transmission target data in response to output data of the model after learning, The learning data includes the driving condition data and the surrounding situation data collected during the specified period as the input data, and includes the judgment results by the judgment function collected during the specified period as the output data of the model.

Images