Driving assistance device and computer program

Abstract

A driving assistance device and a computer program are provided that can efficiently execute calculation processing in driving assistance using big data. [Solution] A driving assistance device (10) is provided with a calculation unit (11) that performs driving assistance according to the traffic risk of the road environment in which the vehicle is located. The calculation unit acquires information about the road environment, determines the attributes of the road environment based on the information, and calculates the traffic risk based on at least one machine learning model out of multiple machine learning models that has learned the features of the attributes.

Description

[Technical Field]

[0001] The present invention relates to a driving assistance device and a computer program for performing driving assistance. [Background technology]

[0002] There is known a technology for calculating the risk of a moving object jumping out from a driver's blind spot at an intersection (see, for example, Patent Document 1). According to this technology, factors that prevent a moving object from entering an intersection are calculated based on the attributes of the road environment where the vehicle is located, and the risk is calculated according to the factors. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent Publication No. 2021-89698 Summary of the Invention [Problem to be solved by the invention]

[0004] In order to realize more advanced driving assistance, the use of big data that includes not only information on road environments but also other information such as pedestrian flow and accident information is being considered. If a determination process using big data is to be performed based on a single machine learning model such as that shown in Patent Document 1, the amount of data collection processing and calculation processing required for the determination may become enormous.

[0005] The present invention aims to provide a driving assistance device and a computer program that can efficiently execute calculation processing in driving assistance using big data. [Means for solving the problem]

[0006] One aspect of the present invention is a driving assistance device that includes a calculation unit that performs driving assistance in accordance with the traffic risk of a road environment in which a vehicle is located, and the calculation unit acquires information about the road environment, determines attributes of the road environment based on the information, and calculates the traffic risk based on at least one machine learning model out of multiple machine learning models that has learned features of the attributes. [Effects of the Invention]

[0007] According to the present invention, it is possible to efficiently execute calculation processing in driving assistance using big data. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a block diagram showing a configuration of a vehicle system according to an embodiment. [Figure 2] FIG. 10 is a diagram illustrating an example of map data used to calculate traffic risks. [Figure 3] 3 is a flowchart showing a processing flow of a driving assistance method. DETAILED DESCRIPTION OF THE INVENTION

[0009] As shown in FIG. 1, a vehicle system S is configured with a vehicle 1 and a server device 20 interconnected by a network W. Hereinafter, the vehicle 1 will also be referred to as the host vehicle. The vehicle 1 is equipped with a driving assistance device 10 that performs driving assistance. The vehicle 1 is equipped with a detection unit 2 that performs detection necessary for the driving assistance. The detection unit 2 is configured to, for example, detect the environment around the vehicle 1 and output the detected value.

[0010] The detection unit 2 is configured with, for example, a camera 2A. The camera 2A captures an image, for example, in the traveling direction of the vehicle 1, and outputs the captured image data to the driving assistance device 10. The detection unit 2 may be configured with one or more cameras 2A so as to capture an image of the environment around the vehicle 1. The detection unit 2 may be provided with not only the camera 2A, but also a lidar device 2B or a radar device 2C that detects objects around the vehicle 1.

[0011] The LIDAR device 2B scans with laser light and measures the reflected light to acquire three-dimensional data of objects around the vehicle 1. The radar device 2C detects objects around the vehicle 1 by emitting radar waves and measuring the reflected waves. The detection unit 2 may be provided with a position sensor 2D that measures the current position of the vehicle 1. The position sensor 2D is configured, for example, with a GPS (Global Positioning System) sensor or the like. The position sensor 2D is used, for example, in a navigation device. The position sensor 2D may also be used in devices other than a navigation device.

[0012] The camera 2A may be configured, for example, as a monocular camera or as a compound camera. The camera 2A captures images of the environment at least in the direction of travel of the vehicle 1 and generates image data including the lane on which the vehicle 1 is traveling. The image data is video data generated based on a predetermined frame rate and can be converted into still image data by dividing the frames. The camera 2A is used for driving assistance control of the vehicle 1. The camera 2A may also be used as a drive recorder.

[0013] The vehicle 1 is equipped with a display unit 3 capable of displaying an image. The display unit 3 is controlled by a driving assistance device 10. The display unit 3 is configured with a display device such as a liquid crystal display or an organic EL (Electro-Luminescence) display. For example, the display unit 3 displays a display image indicating notification content when driving assistance is executed.

[0014] The display unit 3 may be configured to display the display contents of a navigation device provided in the vehicle 1. The display unit 3 may be configured as a touch panel. The display unit 3 may be configured as an input unit that accepts input operations input by a user. In this case, the display unit 3 may display a display image for accepting the input operations. The display unit 3 may be realized by communicating with a mobile terminal device such as a smartphone carried by the user.

[0015] The vehicle 1 is equipped with a steering unit 4 for controlling the direction of travel. The steering unit 4 is composed of a steering device that accepts steering operations from the driver, and a steering device that changes the direction of the wheels of the vehicle 1 based on the steering operation of the steering device. The steering unit 4 is operated in cooperation between the driver and the driving assistance device 10. When the vehicle 1 is a manually driven vehicle, some or all of the operation of the steering unit 4 is controlled by the driving assistance device 10. When the vehicle 1 is an autonomous vehicle, the operation of the steering unit 4 is automatically controlled by the driving assistance device 10. The steering unit 4 is controlled by the driving assistance device 10, and steering operations are performed when lane departure prevention control is executed.

[0016] The vehicle 1 is equipped with a drive unit 5 that serves as a drive source for traveling. The drive unit 5 may be configured with an internal combustion engine or an electric motor. The drive unit 5 may be configured with a hybrid device that combines an internal combustion engine and an electric motor. If the vehicle 1 is a manually driven vehicle, the drive unit 5 is controlled based on operation by the driver, and under predetermined conditions, a driving assistance device 10 executes driving assistance control that assists the driver's operation. If the vehicle 1 is an autonomous vehicle, the drive unit 5 is controlled by the driving assistance device 10.

[0017] The vehicle 1 is equipped with a braking unit 6 for slowing down the vehicle speed. The braking unit 6 is configured, for example, by a brake device. The braking unit 6 is operated in cooperation between the driver and the driving assistance device 10. When the vehicle 1 is a manually driven vehicle, the operation of the braking unit 6 is partially or entirely controlled by the driving assistance device 10. When the vehicle 1 is an autonomous vehicle, the operation of the braking unit 6 is automatically controlled by the driving assistance device 10. The braking unit 6 is controlled by the driving assistance device 10, and braking operations such as automatic braking are performed. When the driving unit 5 is configured by an electric motor, the braking unit 6 may be configured by the driving unit 5. In this case, the driving unit 5 may be configured to decelerate the vehicle 1 by regenerating power based on the deceleration energy of the vehicle 1.

[0018] The vehicle 1 includes a communication unit 7 that is communicatively connected to the network W. The communication unit 7 is configured with a communication device that is capable of wireless communication. The communication unit 7 communicatively connects the driving assistance device 10 and the server device 20 to each other via the network W.

[0019] The driving assistance device 10 includes a calculation unit 11 that performs calculations and control related to driving assistance for the vehicle 1, and a storage unit 12 that stores information. The calculation unit 11 is configured with at least one hardware processor such as a CPU (Central Processing Unit). The calculation unit 11 performs driving assistance according to the traffic risk of the road environment in which the vehicle is located, as described below. The storage unit 12 is configured with a non-transitory storage medium such as a hard disk drive (HDD) or a solid state disk (SSD). The storage unit 12 may also store map data used in the navigation device.

[0020] The storage unit 12 stores data of the detected values ​​output from the detection unit 2. The detected value data may be stored for a predetermined period of time and then updated with new detected value data. The calculation unit 11 is configured with at least one hardware processor such as a CPU (Central Processing Unit). The calculation unit 11 may be realized by hardware (including circuitry) such as an LSI (Large Scale Integration), an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or a GPU (Graphics Processing Unit), or may be realized by a combination of software and hardware.

[0021] The storage unit 12 is configured with a non-transitory storage medium such as a hard disk drive (HDD) or a solid state disk (SSD). The storage unit 12 stores data and programs necessary for calculation and control. The computer programs may be stored in advance in a storage device such as the HDD or flash memory of the storage unit 12, or may be stored in a removable storage medium and installed in the HDD or flash memory of the storage unit 12 by inserting the storage medium into a drive device.

[0022] The server device 20 is configured to communicate with the driving assistance device 10 and provide information necessary for driving assistance. The server device 20 may be integrated into the driving assistance device 10. The server device 20 includes a calculation unit 21 that executes calculations necessary for communication and information provision. The calculation unit 21 is configured with a hardware processor such as a CPU. The server device 20 includes a storage unit 22 that stores information. The storage unit 22 stores data and programs necessary for the calculations of the driving assistance device 10.

[0023] The storage unit 22 stores, for example, map data and a map data set that combines the map data and data on the current position of the vehicle. The storage unit 22 stores big data such as traffic accident data, vehicle data, weather and disaster data, people flow data, and land use classification data related to the map data. The map data set includes feature quantities that are quantified land attributes and related information. The attribute data includes feature quantities that are quantified information such as land use types (commercial areas, residential areas, business areas, green spaces, parks, etc.), population according to date and time, and mobile population. The information related to land attributes and how to use the map data will be described later.

[0024] The server device 20 includes a communication unit 23 that is communicatively connected to the network W. The communication unit 23 is configured by a communication device that is capable of wireless communication. The communication unit 23 communicatively connects the server device 20 and the driving assistance device 10 to each other via the network W.

[0025] The following describes driving assistance performed while the vehicle 1 is traveling. The calculation unit 11 is configured to perform driving assistance according to the traffic risk of the road environment on which the vehicle is located. The calculation unit 11 acquires detection values ​​from the detection unit 2 while the vehicle 1 is traveling. The calculation unit 11 acquires detection values ​​for determining the road environment, such as position data, imaging data, radar data, and lidar data. The calculation unit 11 searches for map data in the server device 20. The calculation unit 11 searches for map data including the current position of the vehicle 1, and acquires information about the road environment on which the vehicle 1 is traveling.

[0026] FIG. 2 shows a map M1 including a road environment in which the vehicle 1 is located. The calculation unit 11 performs driving assistance according to, for example, the traffic risk of the road environment in which the vehicle is located. The calculation unit 11 is configured to perform machine learning such as a neural network in advance, recognize the attributes of the road environment based on the detected values, and perform driving assistance according to the traffic risk of the road environment. The calculation unit 11 analyzes data included in the map data and determines the attributes of the road environment.

[0027] For example, if the road environment is included in a commercial or business site, the calculation unit 11 determines that the road environment is a first road environment E1 of a first attribute. For example, if the road environment is included in a public facility, the calculation unit 11 determines that the road environment is a second road environment E2 of a second attribute. For example, if the road environment is included in a park or green space, the calculation unit 11 determines that the road environment is a third road environment E3 of a third attribute.

[0028] The calculation unit 11 calculates traffic risk based on at least one machine learning model that has learned the association between attribute features and traffic risk, among multiple machine learning models set according to the attributes of the road environment. The machine learning model is set to calculate traffic risk according to the attributes using big data such as traffic accident data, vehicle data, weather and disaster data, pedestrian flow data, and land use classification data as training data. The calculation unit 11 executes driving assistance according to the calculated level of traffic risk. The calculation unit 11 adjusts the content of driving assistance, such as speed adjustment, automatic braking, and avoidance assistance, according to the level of traffic risk.

[0029] For example, when the road environment has a first attribute including a commercial or business location, the calculation unit 11 calculates the traffic risk using a first machine learning model according to the first attribute. The first machine learning model is a machine learning model set to calculate the traffic risk according to a commercial or business location whose population and traffic volume change according to date and time feature values, such as weekdays, holidays, and time periods. The first machine learning model is used to calculate the traffic risk that reflects the first feature value indicating the date and time. The calculation unit 11 adjusts the content of the driving assistance according to the date and time based on the traffic risk that reflects the first feature value. The calculation unit 11 may calculate the degree of traffic risk by weighting according to an event that will be held on a predetermined date and time based on the first machine learning model.

[0030] When the road environment has a second attribute including public facilities, the calculation unit 11 calculates a traffic risk that reflects a second feature amount indicating a moving population based on a second machine learning model according to the second attribute. The second machine learning model is, for example, a machine learning model set to calculate a traffic risk according to public facilities whose moving population changes depending on the date, time, and time of day. The second machine learning model is used to calculate a traffic risk that reflects the second feature amount indicating a moving population. The calculation unit 11 adjusts the content of driving assistance according to the moving population based on the traffic risk that reflects the second feature amount. The calculation unit 11 may also calculate the degree of traffic risk by weighting according to an event that will be held on a specified date and time based on the second machine learning model.

[0031] When the road environment has a third attribute including parks and green spaces, the calculation unit 11 calculates a traffic risk that reflects a third feature value indicating the surrounding population based on a third learning model corresponding to the third attribute. The third machine learning model is, for example, a machine learning model configured to calculate the traffic risk based on parks and green spaces whose surrounding population changes depending on the date, time, and time of day. The third machine learning model is used to calculate the traffic risk that reflects the third feature value indicating the surrounding population. The calculation unit 11 adjusts the content of driving assistance according to the surrounding population based on the traffic risk that reflects the third feature value. The calculation unit 11 may also calculate the degree of traffic risk by weighting the second machine learning model according to environmental changes that occur at a predetermined time, such as flowering information. The above attributes and machine learning models are examples, and the calculation unit 11 may also calculate traffic risks that reflect feature values ​​according to other road environment attributes.

[0032] 3 shows the processing flow of a driving assistance method executed by the driving assistance device 10. The driving assistance method is executed based on a computer program installed in a computer mounted on the driving assistance device that executes driving assistance according to the traffic risk of the road environment in which the vehicle 1 is located. The computer program causes the driving assistance device 10 to execute the following processes.

[0033] The calculation unit 11 acquires a detection value including information about the road environment in which the vehicle 1 is located from the detection unit 2 (S100). The calculation unit 11 determines whether the road environment has a first attribute based on the acquired information (S102). If the road environment has the first attribute, the calculation unit 11 calculates a traffic risk based on a first machine learning model that has learned feature quantities of the first attribute (S104). The calculation unit 11 executes driving assistance according to the traffic risk (S106).

[0034] If the road environment is not the first attribute in S102, the calculation unit 11 determines whether the road environment is the second attribute based on the acquired information (S108). If the road environment is the second attribute, the calculation unit 11 calculates a traffic risk based on a second machine learning model that has learned the feature amount of the second attribute (S110). The calculation unit 11 executes driving assistance according to the traffic risk (S106).

[0035] If the road environment is not the second attribute in S108, the calculation unit 11 determines whether the road environment is the third attribute based on the acquired information (S112). If the road environment is the third attribute, the calculation unit 11 calculates a traffic risk based on a third machine learning model that has learned the feature amount of the third attribute (S114). The calculation unit 11 executes driving assistance according to the traffic risk (S106).

[0036] As described above, the driving assistance device 10 can efficiently perform calculation processing in driving assistance using big data by calculating traffic risks based on at least one machine learning model that has learned attribute features from among multiple machine learning models. The driving assistance device 10 can perform driving assistance suited to the road environment by calculating traffic risks using a machine learning model suited to the road environment. The driving assistance device 10 can perform driving assistance suited to the road environment by calculating traffic risks based on features of the road environment such as the date and time, moving population, and nearby staying population.

[0037] In the above-described embodiment, the computer program executed in each component of the driving assistance device 10 may be provided in a form recorded on a computer-readable, portable, non-transitory recording medium such as a semiconductor memory, a magnetic recording medium, or an optical recording medium. [Explanation of symbols]

[0038] 1 vehicle, 2 detection unit, 2A camera, 2B lidar device, 2C radar device, 2D position sensor, 3 display unit, 4 steering unit, 5 drive unit, 6 braking unit, 7 communication unit, 10 driving assistance device, 11 calculation unit, 12 memory unit, 20 server device, 21 calculation unit, 22 memory unit, 23 communication unit, E1 first road environment, E2 second road environment, E3 third road environment, M1 map, S vehicle system, W network

Claims

1. a calculation unit that executes driving support according to traffic risks of a road environment where the vehicle is located; The calculation unit acquiring information about the road environment; determining attributes of the road environment based on the information; calculating the traffic risk based on at least one machine learning model among a plurality of machine learning models that has learned the feature amount of the attribute; Driving assistance device.

2. The calculation unit When the road environment has a first attribute including a commercial land or a business land, the traffic risk is calculated based on a first machine learning model according to the first attribute, the first feature amount indicating a date and time being reflected; The driving assistance device according to claim 1 .

3. The calculation unit When the road environment has a second attribute including public facilities, the traffic risk is calculated based on a second machine learning model according to the second attribute, the second feature amount indicating a moving population being reflected therein. The driving assistance device according to claim 1 .

4. The calculation unit When the road environment has a third attribute including a park or green space, the traffic risk is calculated based on a third learning model corresponding to the third attribute, the third feature amount indicating a surrounding population being visited. The driving assistance device according to claim 1 .

5. A computer program installed on a computer mounted on a driving assistance device that executes driving assistance according to traffic risks of a road environment in which a vehicle is present, acquiring information about the road environment; determining attributes of the road environment based on the information; causing the computer to execute a process of calculating the traffic risk based on at least one machine learning model that has learned feature quantities of the attribute from among a plurality of machine learning models; Computer program.

Images