Assistance system

Abstract

An assistance system (100, 200, 300) comprises: a position information acquisition unit (11, 211); a sensor information acquisition unit (12, 212); an identification unit (14, 214) whereby an environmental variable (Loc) that is hierarchically determined by classifying information relating to the external environment of a vehicle at different abstraction levels is identified on the basis of position information and sensor information; a data collection unit (219) that collects travel data; a data storage unit that classifies and stores the travel data on the basis of the environmental variable identified by the identification unit; and a generation unit that references the travel data classified by the data storage unit on the basis of information from any of the hierarchical levels, performs machine learning, and generates a rule or an inference model for determining a function to be executed by the vehicle and the control content of the function.

Description

Support system Cross-reference to related applications

[0001] This application is based on Japanese Application No. 2025-004706 filed on January 14, 2025, the contents of which are incorporated herein by reference.

[0002] This disclosure relates to a support system.

[0003] In recent years, systems have been developed that perform image recognition processing on camera images and perform various processes based on the results of the image recognition processing, such as processes related to driving support. When performing this image recognition processing, there are some that execute the image recognition processing using parameters corresponding to the driving area in consideration of the characteristics of the driving area, such as differences in signs for each country. Thereby, image recognition processing corresponding to the characteristics of the driving area can be implemented. Such a technique is described in, for example, Patent Document 1.

[0004] Patent No. 6044556

[0005] By the way, the method for specifying the driving area in the prior art was position specification by GPS or self-reporting, and was limited to a rough classification. That is, it did not perform image recognition processing in consideration of detailed circumstances such as the driving environment of the vehicle during driving and the driving scene. Also, it was not done to select and execute an appropriate driving support function according to the driving area. For this reason, it was not possible to perform optimal driving support corresponding to the sequentially changing external situations.

[0006] This disclosure has been made in view of the above circumstances, and the main object is to provide a support system capable of performing appropriate driving support corresponding to sequentially changing external situations.

[0007] A first support system for solving the above problems is a support system for assisting the driving of a vehicle, comprising: a location information acquisition unit that acquires location information relating to the position of the vehicle; a sensor information acquisition unit that acquires sensor information relating to the surrounding conditions of the vehicle; an identification unit that identifies environment variables which are defined in a hierarchical manner by classifying information relating to the external environment of the vehicle at different levels of abstraction based on the location information acquired by the location information acquisition unit and the sensor information acquired by the sensor information acquisition unit; a data collection unit that collects driving data which includes at least one of the location information and the sensor information; a data storage unit that classifies and stores the driving data based on the environment variables identified by the identification unit; and a generation unit that performs machine learning by referring to the driving data classified by the data storage unit based on information at one of the levels, and generates rules or inference models for determining the functions that the vehicle should perform and the control content of those functions.

[0008] This allows for appropriate driving assistance to be provided in response to constantly changing external conditions.

[0009] A second support system for solving the above problems is a support system equipped with a vehicle control device for controlling a vehicle, wherein the vehicle control device comprises: a location information acquisition unit for acquiring location information relating to the position of the vehicle; a sensor information acquisition unit for acquiring sensor information relating to the surrounding conditions of the vehicle; a calculation unit for performing one or more functions relating to vehicle control; a control content determination unit for determining the functions to be performed by the calculation unit and the control content of those functions; and an identification unit for identifying environment variables based on the location information acquired by the location information acquisition unit and the sensor information acquired by the sensor information acquisition unit, wherein the environment variables are defined in a hierarchical manner by classifying information relating to the external environment of the vehicle at different levels of abstraction, and the control content determination unit determines the functions to be performed by the calculation unit and the control content of those functions based on the environment variables.

[0010] This allows for appropriate driving assistance to be provided in response to constantly changing external conditions.

[0011] The above-mentioned objectives and other objectives, features and advantages of this disclosure will become clearer from the following detailed description with reference to the attached drawings. The drawings are as follows: Figure 1 is a schematic diagram of the support system; Figure 2 is a block diagram showing the functions of the vehicle control device; Figure 3 is a diagram showing the data structure of environment variables; Figure 4 is a diagram showing an example of environment variables; Figure 5 is a block diagram showing the functions of a modified vehicle control device; Figure 6 is a block diagram showing the functions of a modified vehicle control device; Figure 7 is a schematic diagram of the support system of the second embodiment; Figure 8 is a block diagram showing the functions of the vehicle control device of the second embodiment; and Figure 9 is a block diagram showing the functions of the server of the second embodiment.

[0012] The embodiments of the support system described herein will be described in detail below with reference to the drawings. In principle, the same or corresponding parts in the drawings will be denoted by the same reference numerals, and their descriptions will not be repeated between each embodiment and its modified form.

[0013] <First Embodiment> Figure 1 shows a support system 100 to which the vehicle control device 10 in this embodiment is applied. The vehicle control device 10 is mounted on a vehicle 101 and performs control and driving support for the vehicle 101.

[0014] As shown in Figure 1, the support system 100 includes a server 102 and is capable of communicating with one or more vehicles 101 via a communication network 103 such as the Internet. In Figure 1, only one vehicle 101 is shown. The vehicle 101 is equipped with a vehicle control device 10, a sensor 20, an actuator 30, a GPS 40, and the like.

[0015] Sensor 20 includes various sensors for measuring the driving conditions of the vehicle 101, such as a vehicle speed sensor 21, an acceleration sensor 22, and a yaw rate sensor 23. The sensors for measuring the driving conditions of the vehicle 101 may include other sensors, and it is not necessary to have any of the vehicle speed sensor 21, acceleration sensor 22, or yaw rate sensor 23.

[0016] Furthermore, the sensor 20 includes various sensors for detecting other vehicles and obstacles, such as a camera 24 and a millimeter-wave radar 25. The sensors for detecting other vehicles and obstacles are not limited to the millimeter-wave radar 25, but may also be laser radar (LiDAR), ultrasonic sensors, etc., and multiple such sensors may be provided in combination. The camera 24 may be a monocular or a compound camera. The camera 24 may capture either still images or videos. The number, position, and type of cameras 24 may be changed as desired. For example, the vehicle may be equipped with a front camera that captures the area in front of the vehicle, a right side camera that captures the area on the right side of the vehicle, a left side camera that captures the area on the left side of the vehicle, and a rear camera that captures the area behind the vehicle.

[0017] Furthermore, the sensor 20 includes various sensors for detecting various operations performed by the driver, such as an accelerator sensor 26 for detecting the amount of accelerator operation by the driver, a brake sensor 27 for detecting the amount of brake operation, and a steering angle sensor 28 for detecting the amount of steering (steering angle) of the steering wheel by the driver.

[0018] These sensors 20 are connected to the vehicle control device 10 wirelessly or via a wire, and the measurement results (or detection results) from these sensors 20 are input to the vehicle control device 10 as sensor information.

[0019] The actuator 30 includes, for example, an actuator for driving the vehicle 101, such as a motor that serves as the main engine of the vehicle 101. It may also include actuators for controlling the behavior of the vehicle 101, such as an actuator for operating the steering wheel, an actuator for operating the brake pedal, and an actuator for operating the accelerator pedal. Furthermore, the actuator 30 may include devices for operating equipment of the vehicle 101, such as a display, speakers, indicators, and headlights. The driving and operation of the vehicle 101 are controlled by this actuator 30.

[0020] GPS 40 is a global positioning system that detects information regarding the longitude and latitude of the vehicle 101 and inputs it to the vehicle control device 10.

[0021] The vehicle control device 10 is mainly composed of a microcomputer equipped with an arithmetic processing unit 10a such as a CPU and a storage unit 10b such as various types of memory. The functions provided by the microcomputer can be provided by software recorded in a physical memory device and the computer that executes it, by software only, by hardware only, or by a combination thereof. For example, when the microcomputer is provided by electronic circuits which are hardware, it can be provided by digital circuits including a large number of logic circuits or by analog circuits. For example, the arithmetic processing unit 10a of the microcomputer executes a program stored in a non-transitory tangible storage medium which is its own storage unit 10b. The program includes, for example, a program that realizes the functions shown in Figure 2. When the program is executed, the method corresponding to the program is executed. The storage unit 10b is, for example, a non-volatile memory. The program stored in the storage unit 10b can be downloaded and updated via a communication network 103 such as the Internet, for example, OTA (Over The Air).

[0022] The vehicle control device 10 is equipped with various driving support functions (application programs) to assist in the driving of the vehicle 101, and these functions control the actuator 30 based on sensor information input from the sensor 20.

[0023] Typical driver assistance functions to support the driving of vehicle 101 include, for example, adaptive cruise control (ACC), forward collision warning (FCW), advanced emergency braking system (AEBS), night vision / pedestrian detection (NV / PD), traffic sign recognition (TSR), lane departure warning (LDW), lane keeping assist system (LKAS), rear cross traffic alert (RCTA), adaptive front lighting system (AFS), and advanced parking assist (APA). Vehicle 101 may be equipped with all of these functions, or only some of them. It may also be equipped with other driver assistance functions. Furthermore, it may be equipped with functions to realize autonomous driving of vehicle 101. These functions are realized when the processing unit 10a executes the driver assistance control program stored in the storage unit 10b.

[0024] Incidentally, it is desirable to change the control content and functions implemented by the driver assistance system depending on the circumstances of the area in which vehicle 101 is traveling and the surrounding conditions during driving. For example, since speed limits and rules differ from country to country and region to region, it is desirable to change the control content and functions implemented to reflect these according to the area in which the vehicle is traveling. Also, if there are schools or other facilities in the area in which the vehicle is traveling, it is desirable to change the control content and functions implemented to ensure safer driving. Furthermore, even within the same city, it is desirable to change the control content and functions implemented depending on whether or not there is traffic congestion. Similarly, it is desirable to change the control content and functions implemented depending on the weather and time of day. Likewise, it is desirable to change the control content and functions implemented depending on whether or not there is a parade, road construction, or emergency vehicles.

[0025] Therefore, in the vehicle control device 10 of this embodiment, the control content and functions to be performed are changed according to the position of the vehicle 101 and the surrounding conditions during driving, as follows. This will be explained in detail below with reference to the drawings.

[0026] As shown in Figure 2, the arithmetic processing unit 10a of the vehicle control device 10 includes the functions of a position information acquisition unit 11, a sensor information acquisition unit 12, a recognition unit 13, an identification unit 14, a planning unit 15, a control content determination unit 16, a selection unit 17, and a vehicle control unit 18. These functions are realized by the arithmetic processing unit 10a when a program (vehicle control program) stored in the storage unit 10b is executed. In the first embodiment, the vehicle control unit 18 corresponds to the arithmetic unit.

[0027] The location information acquisition unit 11 acquires location information of the vehicle 101 from the GPS 40. Specifically, the location information acquisition unit 11 acquires information regarding the longitude and latitude of the location of the vehicle 101. The location information acquisition unit 11 may also acquire map information along with the location information. The map information can be acquired from the storage unit 10b of the vehicle control device 10. The map information may also be acquired from an external server via the communication network 103 or from an in-vehicle navigation system. The location information acquisition unit 11 inputs the acquired location information to the recognition unit 13 and the identification unit 14.

[0028] The sensor information acquisition unit 12 acquires sensor information from the sensor 20. The sensor information acquisition unit 12 then inputs the acquired sensor information to the recognition unit 13 and the identification unit 14.

[0029] The recognition unit 13 recognizes targets and road information from sensor information. For example, the recognition unit 13 recognizes the position of a target based on camera images captured by the camera 24 as sensor information, or detection results from the millimeter-wave radar 25, etc. Similarly, the recognition unit 13 recognizes road information such as road shape and road condition based on camera images captured by the camera 24 as sensor information, or detection results from the millimeter-wave radar 25, etc. The recognition unit 13 inputs the recognition results (target and road information, etc.) to the planning unit 15 and the control content determination unit 16.

[0030] The identification unit 14 identifies the environment variable Loc based on sensor information and location information. The environment variable Loc is a variable that indicates the external environment of the vehicle 101 at the present time, and in this embodiment, it is a variable that indicates the driving environment of the vehicle 101 while driving. The environment variable Loc contains static information and dynamic information. Static information is information that does not change (or does not change easily) while driving, such as the region where the vehicle 101 is located (country, region, etc.), the shape of the road being driven on, and information about facilities in the surrounding area. As illustrated, this information does not change much in, for example, 24 hours (1 day) or 1 week. On the other hand, dynamic information is information that can change sequentially while driving, such as the weather, time of day, and traffic volume while driving. As illustrated, this information can change, for example, on an hourly basis.

[0031] The environment variable Loc has a hierarchical data structure, and is hierarchically organized by classifying the driving environment at different levels of abstraction. The data structure of the environment variable Loc will be explained in detail. The environment variable Loc employs a tree structure, with information that is less likely to change over time classified at higher levels, and information that is more likely to change over time classified at lower levels. For example, the environment variable Loc is hierarchically organized into static information and dynamic information. In the environment variable Loc, static information, which is less likely to change over time, is classified at higher levels, and dynamic information is classified at lower levels. Furthermore, in the environment variable Loc of this embodiment, static information is further classified into multiple levels, with the least likely to change information being classified at the higher levels of the higher levels of the static information. Similarly, in the environment variable Loc of this embodiment, dynamic information is further classified into multiple levels, with the most likely to change information being classified at the lower levels of the lower levels of the dynamic information.

[0032] To explain in more detail, as shown in Figure 3, the environment variable Loc in this embodiment classifies information into five levels. The first level is the highest level, followed by the second, third, fourth, and fifth levels in descending order. The first level contains information indicating the country where the vehicle 101 is located (e.g., Japan, South Korea, the United Kingdom), and the second level contains information indicating the region of the country where the vehicle 101 is located (e.g., local or prefectural name in the case of Japan). The third level contains information related to roads (e.g., multi-lane straight roads, single-lane straight roads, intersections, three-way intersections, T-junctions, merging points, branching points, curved roads, overpasses, bridges, tunnels, etc.). The fourth level contains information specific to the location where the vehicle is traveling. Location-specific information includes, for example, information about events such as festivals, information indicating the presence of schools nearby, and information indicating the possibility of wild animals such as wild boars appearing. The fifth level contains information about the weather, time of day, and whether or not there is traffic congestion. Weather conditions are classified into categories such as clear skies, sunny, cloudy, rainy, foggy, sleet, snow, and thunderstorms. Time of day is classified into categories such as morning, noon, evening, and night.

[0033] For example, as shown in Figure 4, the environment variable Loc is structured as follows: the first level is "Japan", the second level is "Aichi", the third level is "Intersection", the fourth level is "School Present", and the fifth level is "Sunny, Daytime, No Traffic". Note that the classification of different lower levels may differ depending on the classification of the parent higher level. For example, if the first level is "America", the second level will be classified as "New York State", "Kentucky State", "California State", etc.

[0034] The identification unit 14 identifies information related to the first, second, and third layers based on the map information stored in the storage unit 10b and the location information. The map information is stored with location information and road information associated with it. Therefore, it is possible to identify information about the road being traveled from the location information. The map information may be obtained from a navigation system installed in the vehicle 101, or from an external server 102 via the communication network 103.

[0035] The identification unit 14 identifies information related to the fourth layer based on the sensor information. For example, the identification unit 14 can input information such as camera images included in the sensor information into a machine learning-prepared inference model (an inference model for identifying environment variables for the fourth layer), such as a deep neural network, to identify information related to the fourth layer. When identifying information related to the fourth layer, the identification unit 14 may also use location information and external information (such as event information) obtained from the communication network 103 in addition to the sensor information.

[0036] Similarly, the identification unit 14 identifies information related to the fifth layer based on the sensor information. In other words, the identification unit 14 inputs information such as camera images included in the sensor information into a machine learning-prepared inference model such as a deep neural network (an inference model for identifying environment variables for the fifth layer) to estimate information related to the fifth layer. When identifying information related to the fifth layer, the identification unit 14 may also use location information and external information obtained from the communication network 103 (weather information, traffic volume information, time information, etc.) in addition to the sensor information.

[0037] Furthermore, it is not always necessary to use an inference model; the identification unit 14 may identify information related to the fourth layer and information related to the fifth layer based on the similarity between the sensor information and the pattern information (dictionary information) stored in the storage unit 10b. Alternatively, identification by an inference model and identification by the similarity of pattern information may be used in combination. For example, information related to the fourth layer may be identified by the similarity of pattern information, and information related to the fifth layer may be identified by an inference model.

[0038] Then, the identification unit 14 identifies the information of each of the first to fifth levels, and then hierarchically organizes them, that is, establishes parent-child relationships, to determine the environment variable Loc in a tree structure. The identification unit 14 inputs the determined environment variable Loc to the control content determination unit 16 and the selection unit 17.

[0039] The planning unit 15 determines one or more driving support functions to be implemented based on the recognition results (objects and road information) input from the recognition unit 13. The planning unit 15 inputs the determined driving support functions to be implemented into the selection unit 17.

[0040] The planning unit 15 determines one or more driver assistance functions to be implemented based on the recognition results and according to predetermined rules. Alternatively, the planning unit 15 may input the recognition results into an inference model (a function determination inference model for determining driver assistance functions) to determine which driver assistance functions to implement. The control content of the driver assistance functions determined by the planning unit 15 is predetermined and cannot be changed by the environment variable Loc. The planning unit 15 notifies the selection unit 17 of the determined one or more driver assistance functions to be implemented. At the same time, it also notifies the selection unit 17 that the control content of the driver assistance functions determined by the planning unit 15 is predetermined.

[0041] The control content determination unit 16 determines one or more driving assistance functions to be implemented and their control contents based on the recognition result input from the recognition unit 13 and the environment variable Loc input from the identification unit 14. For example, the control content determination unit 16 inputs the recognition result and the environment variable Loc into a machine learning-prepared inference model such as a deep neural network (a control content determination inference model for determining driving assistance functions and their control contents) to determine one or more driving assistance functions to be implemented and their control contents. Specifically, the control contents are control parameters for each function, such as speed, acceleration (or deceleration), and steering angle. The control content determination unit 16 notifies (inputs) the determined driving assistance functions and their control contents to the selection unit 17.

[0042] The selection unit 17 selects a driving support function from among the one or more driving support functions input from the planning unit 15 and the control content determination unit 16 to be actually implemented by the vehicle control unit 18. In this case, the selection unit 17 may change the selected driving support function and its control content according to the environment variable Loc. The one or more driving support functions input from the planning unit 15 and the control content determination unit 16 are those that have been determined by the planning unit 15 or the control content determination unit 16 as driving support functions to be actually implemented, and are hereinafter simply referred to as the driving support functions that are options.

[0043] The selection unit 17 selects a driving support function from the available driving support functions according to a pre-determined rule. In this case, the environmental variable Loc may or may not be used. For example, when the available driving support functions are defined by rules for each country, the selection unit 17 selects a driving support function from the available driving support functions based on the information related to the first layer of the environmental variable Loc. Also, when the available driving support functions are defined by rules according to the weather, the selection unit 17 selects a driving support function from the available driving support functions based on the information related to the fifth layer of the environmental variable Loc.

[0044] Also, when different control contents are notified from the planning unit 15 and the control content determination unit 16 for the selected driving support function, the selection unit 17 selects which of these control contents to use to implement the driving support function.

[0045] For example, when the follow-up driving function is selected from the options and different control contents of the follow-up driving function are notified from the planning unit 15 and the control content determination unit 16, the selection unit 17 determines which control content of the follow-up driving function is appropriate and selects the control content of the follow-up driving function.

[0046] At that time, the most appropriate control content may or may not be selected using the environmental variable Loc. For example, when the information related to the fifth layer is "sunny, daytime, no traffic jam", a pre-determined control content (initial setting content) may be selected according to the rules, and when it is "rainy, night-time, no traffic jam", the control content determined by the control content determination unit 16 according to the rules may be selected. Also, based on the environmental variable Loc, control parameters may be weighted (e.g., weighted average) to determine the driving support function to be selected.

[0047] The vehicle control unit 18 controls an actuator 30 that operates a steering wheel, an accelerator pedal, etc., so as to implement the driving support function selected by the selection unit 17 with the control content selected by the selection unit 17.

[0048] Hereinafter, the effects in the first embodiment will be described.

[0049] The identification unit 14 determines the environment variable Loc based on location information and sensor information regarding the surrounding conditions of the vehicle 101 during driving. The control content determination unit 16 then changes the driving assistance function that the vehicle control unit 18 should perform and the control content of that driving assistance function based on the environment variable Loc. This enables appropriate driving assistance to be provided in response to constantly changing external conditions.

[0050] The environment variable Loc is defined hierarchically by classifying information about the driving environment at different levels of abstraction, with lower-level information being more prone to change over time compared to higher-level information. The identification unit 14 identifies higher-level information (e.g., the first and second levels) in the environment variable Loc based on location information, and identifies lower-level information (e.g., the fifth level) based on sensor information. The control content determination unit 16 then determines the driving assistance function that the vehicle control unit 18 should implement and the control content of that driving assistance function based on the environment variable Loc. This enables appropriate driving assistance in response to continuously changing external conditions.

[0051] Furthermore, the environment variable Loc is defined in a hierarchical manner by classifying information about the driving environment at different levels of abstraction, with the information at lower levels being more prone to change over time compared to the information at higher levels. Therefore, the determination method by the control content determination unit 16 and the selection method by the selection unit 17 can be changed according to the information at each level, allowing for the determination and selection of more appropriate driving assistance functions.

[0052] The identification unit 14 determines the information of the intermediate level (for example, the fourth level) in the environment variable Loc based on location information and sensor information. This makes it possible to appropriately classify information that cannot be identified (or is difficult to identify) using only location information or only sensor information.

[0053] The identification unit 14 inputs sensor information into a machine learning-based inference model (an inference model for identifying environment variables, hereinafter simply referred to as the first inference model) to identify information at a lower level (fifth level) of the environment variable Loc. This eliminates the need to store a large amount of pattern information and determine similarity.

[0054] Furthermore, when the identification unit 14 identifies lower-level information from the environment variable Loc based on the similarity between the sensor information and the pattern information stored in the storage unit 10b, it has the effect of eliminating the need to create a first inference model using machine learning.

[0055] The control content determination unit 16 inputs the environment variable Loc into a machine learning-based inference model (control content determination inference model, hereinafter simply referred to as the second inference model) to determine the driving assistance function to be implemented and the control content of that driving assistance function. This eliminates the need to define rules. This is particularly effective when the number of environment variables Loc is large.

[0056] <Modification of the First Embodiment> In the first embodiment described above, the control content determination unit 16 may directly input location information and sensor information to determine one or more driving support functions to be performed and their control contents, as shown in Figure 5. More specifically, the location information and sensor information may be input to a machine learning-prepared inference model such as a deep neural network to determine (estimate) one or more driving support functions to be performed and their control contents.

[0057] In the first embodiment described above, the control content determination unit 16 may determine one or more driving support functions to be implemented and their control contents, in accordance with predetermined rules, based on the recognition result input from the recognition unit 13 and the environment variable Loc input from the identification unit 14.

[0058] In the first embodiment described above, the control content determination unit 16 may determine one or more driving support functions to be implemented and their control contents in multiple stages based on the environment variable Loc input from the identification unit 14. For example, one or more driving support functions to be implemented may be determined based on the higher-level information of the environment variable Loc, and then the control contents of the determined driving support functions may be determined based on the lower-level information of the environment variable Loc.

[0059] The decision-making process can be carried out according to predetermined rules or using a pre-trained inference model. Furthermore, the decision-making process can be divided into two or more stages. The decision-making method can also differ at each stage; for example, the first stage might involve following rules, while the next stage uses an inference model. Dividing the decision-making process into multiple stages makes it easier to create rules and inference models.

[0060] For example, as shown in Figure 6, the control content determination unit 16 first identifies a lower rule from among lower rules 1 to 3 according to a higher-level rule, based on the information of the first and second levels of the environment variable Loc. In this case, different lower rules 1 to 3 are created for each region, and the lower rules 1 to 3 are identified based on the information of the first and second levels according to the higher-level rule. Then, the control content determination unit 16 may determine one or more driving support functions to be implemented and their control contents according to the identified lower rules 1 to 3, based on the information of the third level and below. In this way, driving support functions with appropriate control contents can be implemented for each region. This makes it easier to create rules and inference models, as only the rules and inference models need to be created for each region.

[0061] In the above embodiment, information that is likely to change over time may be classified as the hierarchy increases, and information that is less likely to change over time may be classified as the hierarchy decreases.

[0062] In the above embodiment, the number of levels in the environment variable Loc may be changed arbitrarily. Furthermore, how information is classified based on which information at each level may also be changed arbitrarily. For example, information related to time of day may be classified at the fifth level, and information related to weather may be classified at the sixth level. It is desirable to classify information that changes little over time as the level goes up (or down), and to classify information that changes a lot over time as the level goes down (or up).

[0063] - When determining the environment variable Loc in the above embodiment, the location of the road (highway, main road, urban area, farm road, mountain road, etc.) may be classified at any level. Also, when determining the environment variable Loc, the driving scene may be classified at any level. Driving scenes include, for example, scenes of overtaking a vehicle in front, scenes of weaving between vehicles, scenes of following a vehicle in front, scenes of a pedestrian crossing in front of vehicle 101 at night, scenes of vehicle 101 merging from an acceleration lane onto the main road on an expressway, scenes of parking or stopping, scenes of waiting at a traffic light, and various other scenes that can be expected when vehicle 101 is driving.

[0064] - In the above embodiment, the recognition unit 13 does not need to be provided. That is, the control content determination unit 16 may determine one or more driving support functions to be implemented and their control contents based on the environment variable Loc.

[0065] - In the above embodiment, the planning unit 15 does not need to be provided. That is, the selection unit 17 may select one or more driving support functions and their control contents from among the functions determined by the control content determination unit 16.

[0066] In the above embodiment, the selection unit 17 may use a machine learning-based inference model to select the function to be performed and its control content. For example, an inference model (selection inference model) may be created that outputs the function to be performed when a choice of driving assistance function and an environment variable Loc are input, and the selection unit 17 may use this inference model to make a selection. The method for selecting the control content is similar.

[0067] <Second Embodiment> The configuration of the support system 100 of the first embodiment may be modified. The support system 200 in the second embodiment will be described below. Components similar to those in the first embodiment will be denoted by the same reference numerals and their descriptions will be omitted.

[0068] Figure 7 shows the support system 200 to which the vehicle control device 210 in this embodiment is applied. The vehicle control device 210 is mounted on the vehicle 101 and performs control and driving support for the vehicle 101.

[0069] As shown in Figure 7, the support system 200 includes a server 202 and is capable of communicating with one or more vehicles 101 via a communication network 103 such as the Internet. In Figure 7, only one vehicle 101 is shown. The vehicle 101 is equipped with a vehicle control device 210, a sensor 20, an actuator 30, a GPS 40, and the like.

[0070] The vehicle control device 210, similar to the first embodiment, is mainly composed of a microcomputer equipped with a processing unit 10a such as a CPU and a storage unit 10b such as various memories, and is capable of performing driving assistance functions.

[0071] Furthermore, the vehicle control device 210 of the second embodiment is configured to allow setting a mode called shadow mode, which allows the driver assistance control program to be verified (hereinafter referred to as the program to be verified) to be operated and its output values ​​checked while the vehicle 101 is running after sale, that is, in an actual use case, in order to verify the performance and safety of the driver assistance control program.

[0072] In shadow mode verification, for example, the program under verification is run in the background, that is, without being involved in vehicle control, and data related to its output values ​​is stored. This data is then sent to the server 102 via the communication network 103 to collect the output value data. Shadow mode is a type of data acquisition mode used for collecting data.

[0073] The following outlines the configuration and functions related to the shadow mode according to this embodiment. The timing for switching to shadow mode is arbitrary, but for example, it may occur when an operation button or touch panel is operated, or when the ignition switch is turned on. The driver assistance control program that actually operates the vehicle 101 will be referred to as the "implementation program" below, in comparison with the program under verification. The program under verification is stored in the storage unit 10b. A dedicated storage device for shadow mode may also be provided.

[0074] During shadow mode, the program under verification is executed, and various functions to support the driving of the vehicle 101 are performed in the background. The driving support functions performed by the execution of the program under verification may be the same as the driving support functions actually implemented by the processing unit 10a, or some may be omitted. In addition, functions other than those actually implemented by the processing unit 10a may be performed. Generally speaking, even if the driving support functions performed by the program under verification are of the same type as the driving support functions performed by the implemented program (for example, the same follow-me driving function), their control content and decision-making methods may differ.

[0075] When the program under verification is running in shadow mode, the arithmetic processing unit 10a receives sensor information from various sensors 20. Based on the received sensor information, the arithmetic processing unit 10a executes various functions of the program under verification and outputs various control signals to operate each actuator 30. These control signals (output results) based on the program under verification are not actually input to the actuators 30, but are stored in the storage unit 10b as data related to the output results. At that time, the arithmetic processing unit 10a also stores the data related to the input sensor information in association with the output results.

[0076] Furthermore, the arithmetic processing unit 10a may store the control signals processed and output based on the implementation program as data related to the output results in the storage unit 10b. In other words, the control signals processed and output based on the implementation program may be stored in order to compare and verify the output results of the implementation program with the output results of the program under verification.

[0077] Furthermore, the arithmetic processing unit 10a may input sensor information (vehicle speed, yaw rate, acceleration amount, accelerator pedal operation amount, brake pedal operation amount, steering angle, etc.) detected when the vehicle 101 is operating, as well as the control signals processed and output based on the implementation program, to the actuator 30, and store this information in the storage unit 10b. In other words, sensor information related to the actual operation of the vehicle 101 for each scene, that is, sensor information necessary to verify the actual operation, may also be stored.

[0078] The vehicle control device 210 uploads the data stored in the storage unit 10b to the server 202 via the communication network 103 at a predetermined transmission timing. The predetermined transmission timing is any timing, but for example, it is the timing when the server 202 issues an upload instruction. Alternatively, it may be the timing when the vehicle 101 is charging, when it is parked, or when the ignition switch is turned off.

[0079] As explained in the first embodiment, the second embodiment also changes the functions and control content to be performed based on the environment variable Loc. Therefore, in order to verify and improve the performance and safety of the driver assistance control program, it is desirable to collect the environment variable Loc along with the data when collecting data. Accordingly, in the second embodiment, the environment variable Loc is collected along with the data. This will be explained in detail below.

[0080] As shown in Figure 8, the arithmetic processing unit 10a of the second embodiment includes, in addition to the functions described in the first embodiment, a function as a location information acquisition unit 211, a function as a sensor information acquisition unit 212, a function as a recognition unit 213, a function as an identification unit 214, a function as a planning unit 215, a function as a control content determination unit 216, a function as a selection unit 217, a function as a verification unit 218, a function as a data collection unit 219, and a function as a transmission unit 220. Although not shown or described, similar to the first embodiment, the arithmetic processing unit 10a includes a function as a location information acquisition unit 11, a function as a sensor information acquisition unit 12, a function as a recognition unit 13, a function as an identification unit 14, a function as a planning unit 15, a function as a control content determination unit 16, a function as a selection unit 17, and a function as a vehicle control unit 18. These functions are realized by the arithmetic processing unit 10a when a program (vehicle control program) stored in the storage unit 10b is executed. In the second embodiment, the verification unit 218 corresponds to the calculation unit.

[0081] The location information acquisition unit 211 has the same functions as the location information acquisition unit 11 and acquires the location information of the vehicle 101. The location information acquisition unit 211 inputs the acquired location information to the recognition unit 213, the identification unit 214, and the data collection unit 219. Since the function of the location information acquisition unit 211 is the same, it may be made common with the location information acquisition unit 11.

[0082] The sensor information acquisition unit 212 has the same functions as the sensor information acquisition unit 12 and acquires sensor information from the sensor 20. The sensor information acquisition unit 212 then inputs the acquired sensor information to the recognition unit 213, the identification unit 214, and the data collection unit 219. Since the sensor information acquisition unit 212 has the same functions, it may be made common with the sensor information acquisition unit 12.

[0083] The recognition unit 213 has the same functions as the recognition unit 13 and recognizes targets and road information from sensor information. The recognition unit 13 inputs the recognition results (targets and road information, etc.) to the planning unit 215, the control content determination unit 216, and the data acquisition unit 219. Note that since the recognition unit 213 has the same functions, it may be made common with the recognition unit 13.

[0084] The identification unit 214 has the same functions as the identification unit 14 and identifies (determines) the environment variable Loc based on sensor information and location information. The identification unit 214 inputs the determined environment variable Loc to the control content determination unit 216, the selection unit 217, and the data acquisition unit 219. Since the identification unit 214 has the same functions, it may be shared with the identification unit 14.

[0085] The planning unit 215 determines which driving assistance function to implement from among the functions to be verified, according to predetermined rules based on the recognition results. The functions to be verified refer to the driving assistance functions realized by the program to be verified. In this respect, the process differs from the planning unit 15 of the first embodiment, which determines a function from among the driving assistance functions that will be reflected in actual vehicle control. The planning unit 215 may also input the recognition results into an inference model to determine which driving assistance function to implement from among the functions to be verified. The planning unit 215 then notifies the selection unit 217 and the data collection unit 219 of the one or more driving assistance functions to be implemented that it has decided to implement. At that time, it also notifies that the control content of the driving assistance function determined by the planning unit 15 is predetermined.

[0086] The control content determination unit 216 determines one or more driving support functions to be implemented and their control contents from among the functions being verified, based on the recognition result input from the recognition unit 213 and the environment variable Loc input from the identification unit 214. The determination method is the same as the method described in the first embodiment or its modified form. Note that the determination method by the control content determination unit 16 (method for determining implemented functions) and the determination method by the control content determination unit 216 (determination method in shadow mode) may be different or the same. The control content determination unit 216 then notifies (inputs) the determined one or more driving support functions to be implemented and their control contents to the selection unit 217 and the data acquisition unit 219.

[0087] The selection unit 217 selects a driving support function to be actually implemented from among the driving support functions input from the planning unit 215 and the control content determination unit 216, in the same manner as the selection unit 17. Also, as in the first embodiment, if different control contents are notified from the planning unit 215 and the control content determination unit 216 for the selected driving support function, the selection unit 217 selects which of those control contents will be used to implement the driving support function.

[0088] Note that the selection method (selection rules, etc.) may differ between the selection unit 217 of the second embodiment and the selection unit 17 of the first embodiment. Also, if the selection method is the same, the selection unit 217 and the selection unit 17 may be made common. The selection unit 217 notifies (inputs) the selected driving assistance function and its control content to the verification unit 218 and the data acquisition unit 219.

[0089] The verification unit 218 operates the program to be verified and has the driving assistance function selected by the selection unit 217 perform the operation with the control content selected by the selection unit 217. The verification unit 218 then inputs the control signals (control signals for the actuator 30) resulting from this processing to the data acquisition unit 219. As mentioned above, these control signals are not input to the actuator 30.

[0090] As shown by the dashed line in Figure 8, the data acquisition unit 219 acquires data (such as location information, sensor information, environment variable Loc, determined driving support function, its control content, and control signals) from the location information acquisition unit 211, sensor information acquisition unit 212, recognition unit 213, identification unit 214, planning unit 215, control content determination unit 216, selection unit 217, and verification unit 218. The data acquisition unit 219 then determines whether the data acquisition start condition has been met based on the acquired data.

[0091] The data collection start conditions may include, for example, a range of one or more parameters (upper limit, lower limit, or both) for when the data collection start conditions are met, as components of the data collection start conditions. Parameters to be judged in the data collection start conditions may include parameters included in sensor information such as vehicle speed, yaw rate, accelerator operation amount, brake operation amount, and steering amount, as well as parameters included in output results such as required torque. Parameters calculated or estimated from sensor information and output results may also be included. For example, relative distance to obstacles (vehicles ahead, pedestrians, etc.), relative speed to obstacles, and collision margin time (TTC: Time-To-Collision), calculated from the recognition results of camera images (number of targets, etc.) and the detection results of the millimeter-wave radar 25, may be set as parameters of the data collection start conditions. Furthermore, the difference between the output results based on the implementation program and the output results of the program under verification may be set as a parameter of the data collection start conditions. For example, the difference between the brake operation amount output by processing based on the implementation program and the brake operation amount output by processing based on the program under verification may be used as a parameter.

[0092] Furthermore, whether or not the scene in which the vehicle 101 is driving (hereinafter simply referred to as "scene") is a predetermined collection scene may also be included as a component of the collection start condition. These scenes are estimated based on sensor information. For example, the vehicle control device 10 can be made to recognize camera images and estimate the scene. More specifically, sensor information such as camera images can be input into a machine learning-prepared inference model such as a deep neural network to estimate the scene. Note that the vehicle control device 210 does not need to perform image recognition; it may be performed by an external device of the vehicle control device 10, such as an image recognition device, and the results may be input as sensor information.

[0093] Furthermore, among the various functions (application programs) based on the program under verification, the functions that have been performed (implemented functions) may be included as components of the data collection start condition. For example, the data collection start condition may be met when the collision damage mitigation braking control function is performed. It can be determined which function has been performed based on the output result (control signal) input from the verification unit 218.

[0094] In this embodiment, the data collection start condition is set by combining the data collection scene, the function to be implemented, and the parameter range (i.e., using an AND condition). For example, in a scene where the vehicle is following a vehicle ahead, the data collection start condition may be met if the vehicle speed (a parameter of the data collection start condition) is equal to or greater than a threshold (50 km / h).

[0095] The combination of components for the data collection start condition may be a collection scene and an implemented function, a collection scene and a parameter range, or an implemented function and a parameter range. Furthermore, a single data collection start condition may include two or more types of implemented functions, and in this case, the implemented functions may be AND conditions or OR conditions. For example, the condition may be that both the forward vehicle proximity warning function and the collision damage mitigation braking control function are implemented, or that either one is implemented. Similarly, the parameter range may be the range of two or more parameters (such as vehicle speed and steering amount).

[0096] Furthermore, the condition for starting data collection is considered met when all the conditions of the constituent elements of the condition for starting data collection are satisfied. For example, if the data collection scene of the condition for starting data collection is "following a vehicle in front", the function for implementing the condition for starting data collection is "follow-me driving function", and the threshold value of the parameter of the condition for starting data collection is "50 km / h or higher", then the condition for starting data collection is considered met when, in the scene of following a vehicle in front, the follow-me driving function is implemented and the vehicle speed (parameter of the condition for starting data collection) is 50 km / h or higher.

[0097] Furthermore, the data collection start condition is not limited to one; multiple data collection start conditions may be set. For example, the first data collection start condition may be a scene in which the vehicle is following a vehicle in front and the vehicle speed is equal to or greater than a first threshold, and the second data collection start condition may be a scene in which a pedestrian crosses in front of the vehicle 101, the distance between the vehicle and the pedestrian is less than or equal to a second threshold, and the amount of brake operation is equal to or greater than a third threshold. Note that no data collection start conditions are set. In other words, the system may be configured to always collect data while the vehicle 101 is running.

[0098] The data acquisition unit 219 stores (collects) the input data (information) in the storage unit 10b when any of the data acquisition start conditions are met. As mentioned above, the data to be collected includes, for example, location information and sensor information. The data to be collected may also include, for example, control signals processed and output based on the program to be verified, or control signals processed and output based on the implementation program. The data to be collected may also include sensor information related to the actual operation of the vehicle 101 for each scene. The data to be collected may also include the difference between the output result based on the implementation program and the output result of the program to be verified. The data to be stored (collected) may be changed depending on the data acquisition start conditions that are met. For example, in a scene where the vehicle is following a vehicle in front, if the data acquisition start condition is met, which is that the vehicle speed is above a first threshold, data related to the vehicle speed may be collected. On the other hand, in a scene where a pedestrian crosses in front of the vehicle 101, if the data acquisition start condition is met, which is that the distance to the pedestrian is below a second threshold and the amount of brake operation is above a third threshold, data related to the amount of brake operation may be collected.

[0099] Furthermore, when the data collection unit 219 collects data, it stores tag information indicating the fulfilled collection start condition and the date and time it was fulfilled, so that the data is grouped together and stored for each fulfilled collection start condition. In other words, sensor information acquired triggered by the collection start condition is grouped together and stored as data, associated with the tag information. At that time, the data collection unit 219 stores the environment variable Loc, which was identified at the time of collection, associated with the data.

[0100] The transmission unit 220 transmits the data collected by the data collection unit 219 and stored in the storage unit 10b to the server 202 via the communication network 103 at a predetermined transmission timing. The predetermined transmission timing is as described above.

[0101] Next, the functions and processing on the server 202 side will be described. Server 202 is equipped with a CPU 300 as a server-side processing unit and a storage 301 as a server-side memory device. The CPU 300 implements various functions by executing server programs stored in the storage 301. For example, the CPU 300 has the function of a data storage unit 311 that stores received data in the storage 301.

[0102] Specifically, as shown in Figure 9, the data storage unit 311 of the server 202 stores the received data in the driving data storage area 321 for storing all data. At that time, the data is stored in association with the environment variable Loc. This makes it possible to search for data from the environment variable Loc. Furthermore, since the environment variable Loc is hierarchical data, it is also possible to search for data based on information at each level. For example, it is possible to classify and search for data by country or region based on information at a higher level (first or second level). It is also possible to classify and search for data by weather or time of day based on information at a lower level (fifth level).

[0103] Furthermore, in this embodiment, the data storage unit 311 of the server 202 classifies the data by country based on the first-level information and stores it in the macro data storage area 322. Also, the data storage unit 311 of the server 202 classifies the data by road type based on the third-level information and stores it in the micro data storage area 323. Note that the macro data storage area 322 and the micro data storage area 323 may store the data itself, or they may store pointers (data indicating the storage location of the data in the driving data storage area 321, etc.).

[0104] Furthermore, the CPU 300 of the server 202 has the function of a generation unit 312 that generates rules and inference models from the data stored in the storage 301. As mentioned above, the rules and inference models are rules and inference models (inference models for determining control content) for determining the driving assistance functions to be performed by the vehicle 101 and the control content thereof.

[0105] The generation unit 312 collects data by country by referring to the data stored in the macro data storage area 322. At this time, data is collected by country regardless of classification at other levels such as road type, weather, and time of day. Then, based on the collected country-specific data, the generation unit 312 generates country-specific rules by performing machine learning or the like. For example, using multiple data acquired in Japan, machine learning is performed to generate rules (or inference models) for determining the functions to be implemented and the control content (e.g., control parameters) of those functions in Japan. In other words, it generates the determination methods (rules and inference models) for the control content determination units 16 and 216 and the selection methods (rules and inference models) for the selection units 17 and 217 to be adopted in Japan.

[0106] Similarly, country-specific rules and inference models (hereinafter simply referred to as macro rules) are generated for other countries. The generation unit 312 then stores the generated macro rules in the macro rule storage area 331 of the storage 301.

[0107] Similarly, the generation unit 312 collects data for each type of road by referring to the data stored in the microdata storage area 323. At this time, data is collected for each type of road regardless of classifications at other levels such as country, region, weather, and time of day. Then, based on the collected data, the generation unit 312 generates rules for each road by performing machine learning or the like.

[0108] For example, using multiple data points acquired at an intersection, machine learning is performed to generate rules (or inference models) for determining the functions to be implemented at the intersection and the control content (e.g., control parameters) of those functions. In other words, methods for determining the control content determination units 16 and 216 to be adopted at the intersection (rules or inference models) and methods for selecting the selection units 17 and 217 (rules or inference models) are generated.

[0109] Similarly, using multiple data points acquired on the curved road, machine learning is performed to generate rules (or inference models) for determining the functions to be implemented and the control content (e.g., control parameters) of those functions on the curved road. In other words, methods for determining the control content determination units 16 and 216 to be adopted on the curved road (rules or inference models) and methods for selecting the selection units 17 and 217 (rules or inference models) are generated.

[0110] Similarly, rules and inference models (hereinafter simply referred to as microrules) are generated for other road types. The generation unit 312 then stores the generated microrules in the microrule storage area 332 of the storage 301.

[0111] Furthermore, the CPU 300 of server 202 functions as an integration unit 313 that integrates macro rules stored in macro rule storage area 331 and micro rules stored in micro rule storage area 332. The integration unit 313 has the function of adjusting which rule to adopt when macro rules and micro rules are adopted (integrated) together and conflict with each other. For example, in Japan (regardless of other conditions such as weather), there is a rule that vehicles should drive at a speed of 60 km / h or less, while in intersections (regardless of the country), there is a rule that vehicles should proceed slowly at a speed of 20 km / h or less. The integration unit 313 adjusts which rule to adopt, or whether to adopt an intermediate value. In this case, it is generally preferable to prioritize the micro rule, but there are cases where this is not advisable, so it is ultimately desirable to make a case-by-case judgment.

[0112] The rules integrated by the integration unit 313 are stored in the integrated rule storage area 333 of the storage 301. The CPU 300 of the server 202 is equipped with a deployment unit 314 that deploys the rules (and / or inference models, hereinafter the same) stored in the integrated rule storage area 333 to each vehicle 101. At a predetermined timing, the deployment unit 314 deploys the rules stored in the integrated rule storage area 333 to each vehicle 101 via the communication network 103. Specifically, the deployment unit 314 deploys the rules stored in the integrated rule storage area 333 to update the verification target program of each vehicle 101. Note that the implementation program may be updated, not just the verification target program. As a result, the determination methods (rules and inference models) of the control content determination units 16 and 216 and the selection methods (rules and inference models) of the selection units 17 and 217 in each vehicle 101 are updated. Note that when deploying the rules, the country or region to which the rules are deployed may be specified.

[0113] The effects of the second embodiment will be described below.

[0114] The data acquisition unit 219 collects data in association with the environment variable Loc input from the identification unit 214 during data acquisition, and the transmission unit 220 transmits the environment variable Loc associated with the data along with the data when transmitting the data. When the server 202 stores the received data in the storage 301, it stores the received environment variable Loc in association with the data. This makes it possible to select appropriate data that corresponds to the continuously changing external conditions of the vehicle 101 based on the environment variable Loc. In other words, it is possible to select data from the collected data by narrowing down the target conditions and performing verification and development.

[0115] For example, to realize a driver assistance function with control content optimized for a country or region, data can be classified by country or region based on higher-level information in the environment variable Loc, and used for rule verification and development, machine learning of inference models, etc. Similarly, to realize a driver assistance function with control content according to the weather, data can be classified by weather based on lower-level information in the environment variable Loc, and used for rule verification and development, machine learning of inference models, etc.

[0116] Specifically, the data storage unit 311 classifies the data by country based on the first-level information and stores it in the macro data storage area 322. Then, the generation unit 312 generates macro rules by referring to the data stored in the macro data storage area 322. This makes it easy to generate macro rules that utilize country-specific data. In doing so, it is possible to generate macro rules that reflect the unique trends of each country without distinguishing between road types, weather conditions, etc.

[0117] Similarly, the data storage unit 311 classifies the data by road type based on the third-level information and stores it in the microdata storage area 323. Then, the generation unit 312 generates microrules by referring to the data stored in the microdata storage area 323. This makes it easy to generate microrules using road-specific data. In doing so, it is possible to generate microrules that have road-specific tendencies without distinguishing between countries, weather, etc.

[0118] The system includes an integration unit 313 that integrates macro rules stored in the macro rule storage area 331 and micro rules stored in the micro rule storage area 332. When integrating macro rules and micro rules, the integration unit 313 adjusts which rule to adopt if there is a conflict between the rules. Because the adjustment and integration are performed before deploying the rules, the burden on the vehicle 101 can be reduced. In addition, a unified set of rules can be deployed, making verification easier.

[0119] The data acquisition unit 219 collects data related to the output result of the function being verified, along with the environment variable Loc, when the function being verified is performed in shadow mode. The transmission unit 220 transmits the environment variable Loc associated with the data along with the data related to the output result of the function being verified when transmitting the data. When the server 202 stores the received data in the storage 301, it stores the received environment variable Loc associated with the data related to the output result of the function being verified. This allows the output results of the function performed in shadow mode to be collected along with the environment variable Loc, enabling appropriate verification according to the situation.

[0120] <Modification of the Second Embodiment> In the second embodiment described above, the arithmetic processing unit 10a (CPU) was made to perform various processes related to shadow mode. However, a separate arithmetic processing unit (CPU) dedicated to shadow mode may be made to perform various processes related to shadow mode. This allows the arithmetic processing unit 10a to concentrate on various processes related to actual vehicle control.

[0121] In the second embodiment described above, when the server 202 saves data to the storage 301, it may mask personally identifiable information (such as facial data, license plate number, or name). Similarly, the processing unit 10a may mask personally identifiable information before transmitting data from the vehicle 101. This prevents unnecessary disputes when the data is used for verification.

[0122] In the second embodiment described above, the environment variable Loc was set on the vehicle 101 side, but the server 202 side may be provided with an identification unit 14 to identify the environment variable Loc and associate it with each data. In this case, the identification method may be changed on the vehicle 101 side and the server 202 side. In other words, the information used to classify information at each level may differ between the vehicle 101 side and the server 202 side. Also, the number of levels of the environment variable Loc may be different on the vehicle 101 side and the server 202 side.

[0123] - In the second embodiment described above, the generation unit 312 may not be provided. Instead of the generation unit 312, the developer may set the rules. Similarly, the integration unit 313 may not be provided. Instead of the integration unit 313, the developer may integrate the rules.

[0124] In the second embodiment described above, the data storage unit 311 may arbitrarily change the method of classifying the data. For example, the data may be classified by region by referring to the second layer, or the data may be classified by weather by referring to the fifth layer. The data storage unit 311 may also classify the data based on information from two or more layers. For example, the data may be classified by country and time of day using information from the first and fifth layers. For example, daytime data and nighttime data may be classified for each country.

[0125] In the second embodiment described above, the server 202 is equipped with the functions of a data storage unit 311, a generation unit 312, an integration unit 313, and an expansion unit 314. However, some or all of these functions may be provided in an external device to the server 202. In other words, an external device may use the data stored in the server 202's storage 301 to perform data classification, rule generation, integration, expansion, etc.

[0126] In the second embodiment described above, the server 202 does not need to save similar data that has duplicate content. Furthermore, the server 202 may delete some or all of the similar data that has duplicate content in order to increase the storage space of the storage 301.

[0127] The control unit and its method described herein may be implemented by a dedicated computer provided by configuring a processor and memory programmed to perform one or more functions embodied by a computer program. Alternatively, the control unit and its method described herein may be implemented by a dedicated computer provided by configuring a processor by one or more dedicated hardware logic circuits. Alternatively, the control unit and its method described herein may be implemented by one or more dedicated computers configured by a combination of a processor and memory programmed to perform one or more functions and a processor configured by one or more hardware logic circuits. Furthermore, the computer program may be stored as instructions executed by the computer on a computer-readable non-transitional tangible recording medium.

[0128] The following is an addendum regarding the technical ideas that can be derived from each of the above embodiments and modifications. [Configuration 1] A support system (100, 200, 300) for assisting the driving of a vehicle (101), comprising: a location information acquisition unit (11, 211) that acquires location information relating to the position of the vehicle; a sensor information acquisition unit (12, 212) that acquires sensor information relating to the surrounding conditions of the vehicle; an identification unit (14, 214) that identifies environment variables (Loc) which are defined in a hierarchical manner by classifying information relating to the external environment of the vehicle at different levels of abstraction based on the location information acquired by the location information acquisition unit and the sensor information acquired by the sensor information acquisition unit; a data acquisition unit (219) that collects driving data including at least one of the location information and the sensor information; a data storage unit that classifies and stores the driving data based on the environment variables identified by the identification unit; and a generation unit that performs machine learning by referring to the driving data classified by the data storage unit based on information at one of the levels, and generates rules or inference models for determining functions that the vehicle should perform and the control content of those functions. [Configuration 2] A support system (100, 200, 300) comprising a vehicle control device (10, 210) for controlling a vehicle (101), wherein the vehicle control device comprises: a location information acquisition unit (11, 211) for acquiring location information relating to the position of the vehicle; a sensor information acquisition unit (12, 212) for acquiring sensor information relating to the surrounding conditions of the vehicle; a calculation unit (18, 218) for performing one or more functions relating to vehicle control; a control content determination unit (16, 216) for determining the function to be performed by the calculation unit and the control content of that function; and an identification unit (14, 214) for identifying an environment variable (Loc) based on the location information acquired by the location information acquisition unit and the sensor information acquired by the sensor information acquisition unit, wherein the environment variable is defined hierarchically by classifying information relating to the external environment of the vehicle at different levels of abstraction, and the control content determination unit determines the function to be performed by the calculation unit and the control content of that function based on the environment variable, in a support system.[Configuration 3] The vehicle control device comprises a selection unit (17, 217) that selects the function to be actually performed in the vehicle and the control content of that function from the content determined by the control content determination unit, wherein the selection unit changes the selected function and control content according to the environment variable, as described in Configuration 2. [Configuration 4] The control content determination unit inputs the environment variable into a machine learning-trained inference model for determining control content to determine the function to be performed and the control content of that function, as described in Configuration 2 or 3. [Configuration 5] The vehicle control device comprises a data collection unit (219) that collects data relating to sensor information acquired by various sensors mounted on the vehicle, and a transmission unit (220) that transmits the data acquired by the data collection unit to a server (202) via a communication network (103), wherein the data collection unit collects the data in association with the environment variable input from the identification unit at the time of data collection, and the transmission unit transmits the environment variable associated with the data along with the data when transmitting the data, as described in any of Configurations 2 to 4. [Configuration 6] The support system according to Configuration 5, wherein the calculation unit is configured to perform a shadow mode in which the function is performed in a manner that does not involve the actual control of the vehicle, and the data collection unit collects data relating to the output result of the function together with the environment variables when the function is performed in the shadow mode using sensor information acquired by various sensors mounted on the vehicle as input values. [Configuration 7] The support system according to any one of Configurations 1 to 6, wherein the environment variables are defined in a hierarchical manner by classifying information relating to the external environment of the vehicle at different levels of abstraction, and the information at the lower levels is more likely to change over time compared to the information at the higher levels, and the identification unit identifies the information at the higher levels in the environment variables based on the location information, and identifies the information at the lower levels in the environment variables based on the sensor information. [Configuration 8] The support system according to any one of Configurations 1 to 7, wherein the identification unit identifies the information at the intermediate levels in the environment variables based on the location information and the sensor information.[Configuration 9] The identification unit inputs the sensor information into a machine learning-trained inference model for identifying environment variables to identify lower-level information among the environment variables, as described in any of Configurations 1 to 8. [Configuration 10] The identification unit identifies lower-level information among the environment variables based on the similarity between the sensor information and pattern information stored in the memory unit, as described in any of Configurations 1 to 9.

[0129] This disclosure is described in accordance with the embodiments, but it is understood that this disclosure is not limited to such embodiments or structures. This disclosure also includes various modifications and variations within the equivalence. In addition, various combinations and forms, as well as other combinations and forms that include only one, more, or fewer of those elements, fall within the scope and concept of this disclosure.

Claims

1. A support system (100, 200, 300) for assisting the driving of a vehicle (101), comprising: a location information acquisition unit (11, 211) for acquiring location information relating to the position of the vehicle; a sensor information acquisition unit (12, 212) for acquiring sensor information relating to the surrounding conditions of the vehicle; an identification unit (14, 214) for identifying environment variables (Loc) that are defined hierarchically by classifying information relating to the external environment of the vehicle at different levels of abstraction based on the location information acquired by the location information acquisition unit and the sensor information acquired by the sensor information acquisition unit; a data acquisition unit (219) for collecting driving data including at least one of the location information and the sensor information; a data storage unit for classifying and storing the driving data based on the environment variables identified by the identification unit; and a generation unit for performing machine learning by referring to the driving data classified by the data storage unit based on information at one of the levels, and generating rules or inference models for determining functions that the vehicle should perform and the control content of those functions.

2. A support system (100, 200, 300) comprising a vehicle control device (10, 210) for controlling a vehicle (101), wherein the vehicle control device comprises: a location information acquisition unit (11, 211) for acquiring location information relating to the position of the vehicle; a sensor information acquisition unit (12, 212) for acquiring sensor information relating to the surrounding conditions of the vehicle; a calculation unit (18, 218) for performing one or more functions relating to vehicle control; a control content determination unit (16, 216) for determining the function to be performed by the calculation unit and the control content of that function; and an identification unit (14, 214) for identifying an environment variable (Loc) based on the location information acquired by the location information acquisition unit and the sensor information acquired by the sensor information acquisition unit, wherein the environment variable is defined hierarchically by classifying information relating to the external environment of the vehicle at different levels of abstraction, and the control content determination unit determines the function to be performed by the calculation unit and the control content of that function based on the environment variable, in a support system.

3. The support system according to claim 2, wherein the vehicle control device comprises a selection unit (17, 217) that selects the function to be actually performed in the vehicle and the control content of that function from the content determined by the control content determination unit, and the selection unit changes the selected function and control content according to the environment variable.

4. The support system according to claim 2, wherein the control content determination unit inputs the environment variables into a machine learning-trained inference model for determining control content to determine the function to be performed and the control content of that function.

5. The support system according to claim 2, wherein the vehicle control device comprises a data acquisition unit (219) that collects data relating to sensor information acquired by various sensors mounted on the vehicle, and a transmission unit (220) that transmits the data acquired by the data acquisition unit to a server (202) via a communication network (103), the data acquisition unit collects the data in association with the environment variables input from the identification unit at the time of data acquisition, and the transmission unit transmits the environment variables associated with the data along with the data when transmitting the data.

6. The support system according to claim 5, wherein the calculation unit is configured to perform a shadow mode in which the function is performed without being involved in the actual control of the vehicle, and the data acquisition unit collects data relating to the output result of the function, together with the environment variables, when the function is performed in the shadow mode using sensor information acquired by various sensors mounted on the vehicle as input values.

7. The support system according to any one of claims 1 to 6, wherein the environment variables are defined in a hierarchical manner by classifying information relating to the external environment of the vehicle at different levels of abstraction, the lower-level information being more prone to change over time compared to the higher-level information, and the identification unit identifies the higher-level information in the environment variables based on the location information, and identifies the lower-level information in the environment variables based on the sensor information.

8. The support system according to any one of claims 1 to 6, wherein the identification unit identifies intermediate hierarchical information in the environment variables based on the location information and the sensor information.

9. The support system according to any one of claims 1 to 6, wherein the identification unit inputs the sensor information into a machine learning-trained inference model for identifying environment variables to identify lower-level information among the environment variables.

10. The support system according to any one of claims 1 to 6, wherein the identification unit identifies lower-level information among the environment variables based on the similarity between the sensor information and the pattern information stored in the storage unit.

Images