JP2023097051A - Driver characteristic estimation system, driver characteristic estimation method and computer program - Google Patents

Abstract

To provide a driver characteristic estimation system capable of improving determination accuracy of a mental state of a driver, a driver characteristic estimation method and a computer program.SOLUTION: A driver characteristic estimation system comprises: a collection unit for collecting vehicle data relating to a vehicle that a driver drives; an acquisition unit for acquiring contract data indicating a contract content based on a portable terminal that the driver carries; and a deviation unit for deviating mental characteristics associating with the contract data or the vehicle data based on a learnt model including a result of performing machine learning on a relation between the contract data and mental characteristics of the driver and a result of performing machine learning on a relation between the vehicle data and the mental characteristics of the driver.SELECTED DRAWING: Figure 3

Description

The present invention relates to a driver characteristic estimation system, a driver characteristic estimation method and a computer program.

The driver's mental state changes according to the driving situation. For example, the driver's mental state includes an irritated mental state, an impatient mental state, and a good mental state.
There is known a technique of switching to a safer control according to the driver's state of mind (see Patent Literature 1, for example).

JP 2013-218491 A

In the technology described above, the psychological state of the driver is determined using facial expressions, brain waves, heart rate, and the like. Therefore, the driver's mental state changes depending on the temporarily obtained facial expression, electroencephalogram, and heart rate.
The driver's state of mind can also be determined by permanent data in addition to temporarily available factors. In order to switch to safer control in a vehicle, it is necessary to improve the determination accuracy of the driver's state of mind.
An object of the present invention is to provide a driver characteristic estimation system, a driver characteristic estimation method, and a computer program capable of improving the accuracy of determination of the driver's mental state.

(1) One aspect of the present invention includes a collection unit that collects vehicle data related to a vehicle driven by a driver, an acquisition unit that acquires contract data indicating contract details based on a mobile terminal owned by the driver, and contract data. Relationship between the acquired contract data or the vehicle data based on a learned model including the result of machine learning of the relationship with the psychological characteristics of the driver and the result of machine learning of the relationship between the vehicle data and the psychological characteristics of the driver. and a derivation unit for deriving psychological characteristics to be used.
(2) In one aspect of the present invention, in the driver characteristic estimation system described in (1) above, a first characteristic quantity that quantitatively represents a characteristic of the contract data acquired by the acquisition unit is generated. a quantity generation unit; and a second feature quantity generation unit for generating a second feature quantity that quantitatively expresses the characteristics of the vehicle data of the vehicle collected by the collection unit, wherein the derivation unit includes a first Based on the learned model including the results of machine learning of the relationship between the feature quantity and the driver's psychological characteristics and the result of machine learning of the relationship between the second feature quantity and the driver's psychological characteristics, the first feature quantity or A psychological characteristic related to the second feature is derived.
(3) In one aspect of the present invention, in the driver characteristic estimation system described in (2) above, the second feature value generation unit converts the vehicle data of the vehicle into the driving habits of the driver of the vehicle and the driving habits of the driver of the vehicle. driving behavior of the driver of the vehicle; a driving habit feature quantity that quantitatively represents the characteristics of the driving habit of the driver of the vehicle; and a driving behavior that quantitatively represents the characteristics of the driving behavior of the driver of the vehicle. Generate features.
(4) In one aspect of the present invention, in the driver characteristic estimation system described in (3) above, the second feature value generation unit divides the driving operation of the driver into a plurality of driving situations, and divides the driving operation into a plurality of driving situations. A second feature amount is generated based on the driving operation of the vehicle driver included in each of the.
(5) In one aspect of the present invention, in the driver characteristic estimation system according to any one of (2) to (4) above, the derivation unit may Then, based on either the result of machine learning of the relationship between the first feature amount and the psychological characteristic of the driver or the result of machine learning of the relationship between the second feature amount and the psychological characteristic of the driver, the first characteristic A psychological characteristic related to the quantity or a psychological characteristic related to the second feature amount is derived.
(6) In one aspect of the present invention, in the driver characteristic estimation system according to any one of (2) to (5) above, for each of the plurality of derived items included in the psychological characteristic, the learned the first feature amount and the second feature amount based on the accuracy of the derived item obtained from the model and the first feature amount and the accuracy of the derived item obtained from the trained model and the second feature amount; is used when deriving derived items.
(7) An aspect of the present invention is the driver characteristic estimation system according to any one of (1) to (6) above, wherein the psychological characteristic is any one of Big 5, driving attitude, and burden sensitivity. or

(8) One aspect of the present invention is a driver characteristic estimation method executed by a computer, comprising: a step of collecting vehicle data relating to a vehicle driven by a driver; Acquisition based on a trained model including a step of acquiring data, the result of machine learning of the relationship between contract data and the driver's psychological characteristics, and the result of machine learning of the relationship between vehicle data and the driver's psychological characteristics and deriving psychological characteristics associated with the contract data or the vehicle data obtained from the driver characteristics estimation method.

(9) One aspect of the present invention is a step of collecting vehicle data relating to a vehicle driven by a driver in a computer, obtaining contract data indicating contract details based on a portable terminal possessed by the driver, and obtaining contract data The obtained contract data or the vehicle data based on a trained model that includes the results of machine learning of the relationship between the driver's psychological characteristics and the results of machine learning of the relationship between the vehicle data and the driver's psychological characteristics. and deriving relevant psychological characteristics.

Advantageous Effects of Invention According to the present invention, it is possible to provide a driver characteristic estimation system, a driver characteristic estimation method, and a computer program capable of improving the determination accuracy of a driver's state of mind.

It is a figure which shows the structural example of the driver characteristic estimation system of embodiment of this invention. It is a figure which shows an example of a process of the driver characteristic estimation system of this embodiment. It is a figure which shows the detail of the driver characteristic estimation system of this embodiment. It is a figure which shows an example of the item contained in contract data. It is a figure which shows an example of the item contained in vehicle data. It is a figure which shows an example of Big5. It is a figure which shows an example of a process of the driver characteristic estimation apparatus of this embodiment. It is a figure which shows an example of a process of the driver characteristic estimation apparatus of this embodiment. It is a figure which shows an example of the learning apparatus of this embodiment. It is a figure which shows an example of a process of the learning apparatus of this embodiment. It is a figure which shows an example of a process of the driver characteristic estimation system of the modified example of embodiment. It is a figure which shows the detail of the driver-characteristics estimation apparatus contained in the driver-characteristics estimation system of the modified example of embodiment. It is a figure which shows an example of a process of the driver characteristic estimation apparatus of the modified example of embodiment. It is a figure which shows an example of a process of the driver characteristic estimation apparatus of the modified example of embodiment. It is a figure which shows an example of a process of the driver characteristic estimation apparatus of the modified example of embodiment. It is a figure which shows an example of a process of the driver characteristic estimation system of the modified example of embodiment. It is a figure which shows an example of the learning apparatus of the modification of embodiment. It is a figure which shows an example of a process of the learning apparatus of the modified example of embodiment. It is a figure which shows an example of a process of the learning apparatus of the modified example of embodiment.

Next, the driver characteristic estimation device, driver characteristic estimation method, and computer program of this embodiment will be described with reference to the drawings. The embodiments described below are merely examples, and embodiments to which the present invention is applied are not limited to the following embodiments. In addition, in all the drawings for describing the embodiments, the same reference numerals are used for those having the same functions, and repeated descriptions are omitted.
In addition, "based on XX" in the present application means "based on at least XX", and includes cases based on other elements in addition to XX. Moreover, the term "based on XX" is not limited to the case of using XX directly, but also includes the case of being based on XX that has undergone calculation or processing. "XX" is an arbitrary element (for example, arbitrary information).

(First embodiment)
(Driver characteristics estimation system)
FIG. 1 is a diagram showing a configuration example of a driver characteristic estimation system according to an embodiment of the invention. In FIG. 1 , the driver characteristic estimation system 1 includes a driver characteristic estimation device 100 . In addition to the driver characteristic estimation device 100, FIG. 1 shows a portable terminal 200 and an on-vehicle device 300 mounted on the vehicle VE. The holder of mobile terminal 200 is driving vehicle VE. Hereinafter, the holder of the mobile terminal 200 is also referred to as a driver.
Driver characteristic estimation device 100, portable terminal 200, and in-vehicle device 300 communicate via base station BS connected to network NW. The network NW includes, for example, the Internet, WAN (Wide Area Network), LAN (Local Area Network), provider equipment, wireless base stations, and the like.

An outline of processing of the driver characteristic estimation system 1 will be described.
FIG. 2 is a diagram showing an example of processing of the driver characteristic estimation system of this embodiment.
The driver characteristic estimation device 100 associates and registers the user ID of the mobile terminal 200 and the identification information of the in-vehicle device (hereinafter referred to as "in-vehicle device ID"). Here, a vehicle ID may be registered instead of the vehicle-mounted device ID. As an example, the case where the vehicle-mounted device ID is registered will be described below. The driver characteristic estimation device 100 registers contract data for user IDs. Here, the contract data of the user ID may be registered in a server different from driver characteristic estimation device 100 . As an example, the case where the contract data of the user ID is registered in the driver characteristic estimation device 100 will be described below.
The in-vehicle device 300 detects operation information of the driver (user) of the vehicle VE and state information of the vehicle VE. The in-vehicle device 300 creates a vehicle data notification including the vehicle data of the vehicle VE including the detected operation information of the vehicle VE and the state information of the vehicle VE and the in-vehicle device ID, and sends the created vehicle data notification to the driver. Transmit to characteristic estimating apparatus 100 (1).

The driver characteristic estimation device 100 receives the vehicle data notification transmitted by the vehicle-mounted device 300 and acquires the vehicle-mounted device ID included in the received vehicle data notification. The driver characteristic estimation device 100 acquires from the storage unit 110 the user ID associated with the acquired vehicle-mounted device ID. Driver characteristic estimation device 100 acquires contract data for the acquired user ID from storage unit 110 .
The driver characteristics estimation device 100 acquires based on a learned model including the results of machine learning of the relationship between the contract data and the psychological characteristics of the driver and the results of machine learning of the relationship between the vehicle data and the psychological characteristics of the driver. Deriving psychological characteristics associated with contract data or acquired vehicle data (2).
Driver characteristic estimation device 100 creates a psychological characteristic notification including the psychological characteristic derivation result, and outputs the created psychological characteristic notification (3). An example of the psychological characteristic derivation result is one of BIG5, driving attitude (DSQ), and burden sensitivity (WSQ).

Hereinafter, each of the driver characteristic estimation device 100, the portable terminal 200, and the in-vehicle device 300 included in the driver characteristic estimation system 1 will be sequentially described.
FIG. 3 is a diagram showing the details of the driver characteristic estimation system of this embodiment.
(Driver characteristic estimation device 100)
Driver characteristic estimation device 100 is realized by a device such as a personal computer, a server, a smart phone, a tablet computer, or an industrial computer. Driver characteristic estimation device 100 includes communication unit 102, acquisition unit 103, creation unit 104, collection unit 105, derivation unit 108, and storage unit 110, for example.
The communication unit 102 is implemented by a communication module. The communication unit 102 communicates with an external communication device via the network NW. The communication unit 102 may communicate using a wireless communication method such as wireless LAN, Bluetooth (registered trademark), or LTE (registered trademark). Also, the communication unit 102 may communicate using a communication method such as a wired LAN.
The communication unit 102 receives the vehicle data notification transmitted by the vehicle-mounted device 300 . The communication unit 102 transmits the psychological characteristic notification output by the creation unit 104 .

The acquisition unit 103 acquires the vehicle data notification received by the communication unit 102, acquires the user ID linked to the vehicle-mounted device ID included in the acquired vehicle data notification from the storage unit 110, and stores the acquired user ID. Contract data is acquired from the storage unit 110 .
The storage unit 110 is implemented by a HDD (Hard Disk Drive), flash memory, RAM (Random Access Memory), ROM (Read Only Memory), or the like.
The storage unit 110 associates and registers the user ID of the mobile terminal 200 and the vehicle-mounted device ID. The storage unit 110 stores the contract data of the user ID and the notification destination list. The notification destination list associates user notification destination information that associates a user ID with a notification destination to a user corresponding to the user ID, and associates an onboard device ID with a notification destination to the onboard device that corresponds to the onboard device ID. In-vehicle device notification destination information. Examples of notification destinations are subscriber identification information such as telephone numbers and e-mail addresses.
FIG. 4 is a diagram showing an example of items included in contract data. As shown in FIG. 4, the contract data includes terminal color, terminal manufacturer, model name, OS, model change/purchase date, communication plan fee, data capacity, whether or not a communication flat rate plan is contracted, smart value (fixed Set discount with line) use or not. In addition, the contract data includes usage of e-book services/monthly usage fee, usage of music services/monthly usage fee, usage of video services/monthly usage fee, usage of e-commerce sites/monthly usage fee, This includes whether or not they are subscribed to Internet banking, whether or not they are subscribed to new electric power companies, and others.

The collection unit 105 acquires the vehicle data notification received by the communication unit 102, collects the vehicle data and the on-vehicle device ID included in the acquired vehicle data notification, and stores them in the storage unit 110 in association with the acquisition time.
FIG. 5 is a diagram showing an example of items included in vehicle data. The vehicle data includes operation information of the driver of the vehicle VE and state information of the vehicle VE. As shown in FIG. 5, the state information of the vehicle VE includes collection method, collection cycle, mileage, vehicle speed, longitudinal acceleration, lateral acceleration, vertical acceleration, engine start/end, and shift position. Further, the operation information of the driver of the vehicle VE includes an accelerator pedal operation amount, a brake pedal operation amount, and a steering angle. Some of the items described above may be included in the operation information of the driver of the vehicle VE and the state information of the vehicle VE.

Derivation unit 108 includes trained model 109 . Derivation unit 108 acquires the contract data of the user ID acquired by acquisition unit 103 . Based on the learned model 109, the derivation unit 108 derives the psychological characteristics related to the acquired contract data or the psychological characteristics related to the vehicle data.
The trained model 109 will be explained. The learned model 109 includes the results of machine learning of the relationship between the contract data and the driver's psychological characteristics, and the results of machine learning of the relationship between the vehicle data and the driver's psychological characteristics. An example of psychological characteristics is one of Big 5, Driving Attitude (DSQ) and Stress Sensitivity (WSQ). The Big 5 can estimate the driver's personality, the DSQ can estimate the driver's driving psychology, and the WSQ can estimate the driver's driving characteristics.

FIG. 6 is a diagram showing an example of Big5. The Big 5 is a quantification of a person's personality on five axes. The Big 5 indicators (five axes) include Openness, Conscientiousness, Extraversion, Agreeableness, and Neuroticism. The Big 5 is being used in various services such as advertising, finance, and human resources.
Driving attitude (DSQ) and strain sensitivity (WSQ) are psychological characteristics related to driving. DSQ (driving attitude) is the attitude, intention, and way of thinking about driving. DSQ indicators include impatient driving tendencies, reluctance to drive, and the like. WSQ (Strain Sensitivity) is what kind of driving strain you feel strongly about. The WSQ index includes distraction to driving, driving posture, and the like.

Whether the trained model 109 is derived using contract data or vehicle data is set for each tendency of the psychological characteristic index.
FIG. 7 is a diagram showing an example of processing of the driver characteristic estimation device of this embodiment. FIG. 7 shows the relationship between an index (derivation target) included in psychological characteristics, the tendency of the index, and usage data used when deriving the tendency of the index (derivation target). Here, as an example, the relationship between the five indices included in Big5, the tendency of each of the five indices, and the usage data used when deriving each of the five indices is shown.
FIG. 7 shows that for intellectual curiosity, vehicle data is used when deriving a high tendency (high), and contract data is used when deriving a low tendency (low). Conscientiousness is shown to use contract data when deriving high propensity and vehicle data when deriving low propensity. For extroversion, it has been shown to use contract data when deriving high propensity and vehicle data when deriving low propensity. Regarding cooperativeness, it is shown that vehicle data is used when deriving a high tendency, and contract data is used when deriving a low tendency. For emotional instability, it is shown to use vehicle data when deriving high propensity and vehicle data when deriving low propensity. Returning to FIG. 3, the description is continued.

The derivation unit 108 inputs vehicle data to the learned model 109 when deriving a high tendency for intellectual curiosity, and acquires information indicating a high tendency for intellectual curiosity output from the learned model 109 . The derivation unit 108 inputs the contract data to the learned model 109 when deriving the tendency of low intellectual curiosity, and acquires the information indicating the tendency of low intellectual curiosity output from the trained model 109 . The derivation unit 108 inputs the contract data to the learned model 109 when deriving a high tendency of sincerity, and acquires information indicating a high tendency of sincerity output from the learned model 109 . The derivation unit 108 inputs the vehicle data to the learned model 109 when deriving the tendency of low conscientiousness, and acquires the information indicating the tendency of low conscientiousness output by the learned model 109 .
The derivation unit 108 inputs the contract data to the learned model 109 when deriving a high extroversion tendency, and acquires information indicating a high extroversion tendency output by the learned model 109 . The derivation unit 108 inputs the vehicle data to the learned model 109 when deriving the tendency of low extroversion, and acquires the information indicating the tendency of low extroversion output from the learned model 109 . The derivation unit 108 inputs the vehicle data to the learned model 109 when deriving the tendency of high cooperativeness, and acquires the information indicating the tendency of high cooperativeness output from the learned model 109 . Derivation unit 108 inputs contract data to learned model 109 when deriving a tendency for low cooperativeness, and acquires information indicating a tendency for low cooperativeness output from learned model 109 .
The derivation unit 108 inputs vehicle data to the learned model 109 when deriving a high tendency for emotional instability, and acquires information indicating a high tendency for emotional instability output from the trained model 109 . The derivation unit 108 inputs the vehicle data to the learned model 109 when deriving a low tendency for emotional instability, and acquires information indicating a low tendency for emotional instability output by the trained model 109 .

The creation unit 104 acquires the information indicating the psychological characteristics acquired by the derivation unit 108, the user ID, and the vehicle-mounted device ID. Specifically, the creation unit 104 acquires information indicating intellectual curiosity, information indicating conscientiousness, information indicating extroversion, information indicating conscientiousness, and information indicating emotional instability.
The creation unit 104 acquires the destination of notification to the portable terminal 200 of the user corresponding to the user ID and the destination of notification to the in-vehicle device 300 corresponding to the in-vehicle device ID from the destination list stored in the storage unit 110. . The creation unit 104 sends the acquired information indicating intellectual curiosity, information indicating conscientiousness, information indicating extroversion, information indicating conscientiousness, and information indicating emotional instability to the mobile terminal 200 of the user as a destination. Create a psychological characteristic notification that The creation unit 104 outputs the created psychological characteristic notification to the communication unit 102 .
In addition, the creation unit 104 includes the information indicating intellectual curiosity, the information indicating conscientiousness, the information indicating extroversion, the information indicating conscientiousness, and the information indicating emotional instability, and includes the acquired information indicating intellectual curiosity, information indicating conscientiousness, and information indicating emotional instability. Create a psychological characteristic notification that The creation unit 104 outputs the created psychological characteristic notification to the communication unit 102 .

Acquisition unit 103, creation unit 104, collection unit 105, and derivation unit 108 are implemented by a hardware processor such as a CPU (Central Processing Unit) executing a computer program (software) stored in storage unit 110. be done.
Further, some or all of these functional units are implemented in hardware such as LSI (Large Scale Integration), ASIC (Application Specific Integrated Circuit), FPGA (Field-Programmable Gate Array), GPU (Graphics Processing Unit), and the like. wear (circuit (including circuit), or by cooperation between software and hardware.

(Portable terminal 200)
The mobile terminal 200 is implemented by a device such as a personal computer, smart phone, tablet computer, or industrial computer. The mobile terminal 200 includes a communication unit 202, an output unit 209, and a storage unit 210, for example.
The communication unit 202 is implemented by a communication module. The communication unit 202 communicates with an external communication device via the network NW. The communication unit 202 may communicate using a wireless communication method such as wireless LAN, Bluetooth (registered trademark), or LTE (registered trademark).
The communication unit 202 receives the psychological characteristic notification transmitted by the driver characteristic estimation device 100 .

The storage unit 210 is implemented by an HDD, flash memory, RAM, ROM, or the like.
The output unit 209 acquires the psychological characteristic notification from the communication unit 202 and outputs the acquired psychological characteristic notification. The output unit 209 may output the psychological characteristic notification by voice, or may output it by displaying it on a display unit (not shown).
The output unit 209 is implemented, for example, by a hardware processor such as a CPU executing a computer program (software) stored in the storage unit 210 . In addition, some or all of these functional units may be realized by hardware (including circuitry) such as LSI, ASIC, FPGA, GPU, etc., or by cooperation of software and hardware. may be implemented.

(Vehicle device 300)
The in-vehicle device 300 is implemented by a device such as a car navigation device, personal computer, smart phone, tablet computer, or industrial computer. The in-vehicle device 300 includes, for example, a communication unit 302, a creation unit 304, a detection unit 306, an output unit 309, and a storage unit 310.
The communication unit 302 is implemented by a communication module. The communication unit 302 communicates with an external communication device via the network NW. The communication unit 302 may communicate using a wireless communication method such as wireless LAN, Bluetooth (registered trademark), or LTE (registered trademark).
Communication unit 302 transmits the vehicle data notification output by creation unit 304 to driver characteristic estimation device 100 . The communication unit 302 receives the psychological characteristic notification transmitted by the driver characteristic estimation device 100 .

The detection unit 306 includes sensors capable of measuring the operation of the vehicle VE and the state of the vehicle VE, such as a speed sensor, an acceleration sensor, and a gyro sensor. The detection unit 306 acquires sensor information obtained by sensing by the sensors as operation information of the vehicle VE and state information of the vehicle VE.
The creation unit 304 acquires the operation information of the vehicle VE and the state information of the vehicle VE from the detection unit 306 . Creation unit 304 creates vehicle data including operation information of vehicle VE and state information of vehicle VE, and notifies vehicle data including the created vehicle data and on-vehicle device ID to driver characteristic estimation device 100 as a destination. create. Creation unit 304 outputs the created vehicle data notification to communication unit 302 .

The output unit 309 acquires the psychological characteristic notification from the communication unit 302 and outputs the acquired psychological characteristic notification. The output unit 309 may output the psychological characteristic notification by voice, or may output it by displaying it on a display unit (not shown).
The creation unit 304 and the output unit 309 are implemented by, for example, a hardware processor such as a CPU executing a computer program (software) stored in the storage unit 310 . In addition, some or all of these functional units may be realized by hardware such as LSI, ASIC, FPGA, GPU (circuit unit; including circuit), or by cooperation of software and hardware. may be implemented.

(Operation of driver characteristic estimation system 1)
FIG. 8 is a diagram showing an example of processing of the driver characteristic estimation system of this embodiment.
(Step S1-1)
In the in-vehicle device 300, the detection unit 306 acquires operation information of the vehicle VE and state information of the vehicle VE obtained by sensing by the sensor.
(Step S2-1)
In the in-vehicle device 300 , the creation unit 304 acquires the operation information of the vehicle VE and the state information of the vehicle VE from the detection unit 306 . The creating unit 304 creates vehicle data including the acquired operation information of the vehicle VE and the state information of the vehicle VE, and sends the vehicle data including the created vehicle data and the vehicle-mounted device ID to the driver characteristic estimation device 100. Create notifications.
(Step S3-1)
In the in-vehicle device 300 , the creation unit 304 outputs the created vehicle data notification to the communication unit 302 . Communication unit 302 transmits the vehicle data notification output by creation unit 304 to driver characteristic estimation device 100 .
In the driver characteristic estimation device 100 , the communication unit 102 receives the vehicle data notification transmitted by the vehicle-mounted device 300 .

(Step S4-1)
In the driver characteristic estimation device 100, the collection unit 105 collects the vehicle data notification received by the communication unit 102, acquires the vehicle data and the vehicle-mounted device ID included in the collected vehicle data notification, and stores the acquisition time in the storage unit 110. be stored in association with The acquisition unit 103 acquires the vehicle-mounted device ID included in the vehicle data notification received by the communication unit 102 and acquires the user ID linked to the acquired vehicle-mounted device ID from the storage unit 110 . Acquisition unit 103 acquires contract data for the acquired user ID from storage unit 110 .
(Step S5-1)
In the driver characteristic estimation device 100 , the derivation unit 108 acquires vehicle data from the collection unit 105 and acquires contract data from the acquisition unit 103 . Based on the learned model 109, the derivation unit 108 derives the psychological characteristics related to the acquired contract data or the psychological characteristics related to the vehicle data.

(Step S6-1)
In the driver characteristic estimation device 100, the creation unit 104 acquires the information indicating the psychological characteristics acquired by the derivation unit 108. FIG. The creating unit 104 creates a psychological characteristic notification addressed to the portable terminal 200, including information indicating the acquired psychological characteristic.
In addition, the creation unit 104 creates a psychological characteristic notification addressed to the in-vehicle device 300 including information indicating the acquired psychological characteristic.
(Step S7-1)
In driver characteristic estimation device 100 , creation unit 104 outputs the created psychological characteristic notification to communication unit 102 . The communication unit 102 transmits the psychological characteristic notification output by the creation unit 104 to the mobile terminal 200 .
In mobile terminal 200 , communication unit 202 receives the psychological characteristic notification transmitted by driver characteristic estimation device 100 .
(Step S8-1)
In the driver characteristic estimation device 100 , the psychological characteristic notification output by the creation unit 104 is transmitted to the vehicle-mounted device 300 .
In the in-vehicle device 300 , the communication unit 302 receives the psychological characteristic notification transmitted by the driver characteristic estimation device 100 .
After that, the process returns to step S1-1, and the same processing as described above is performed at a predetermined cycle.

In the above-described embodiment, as an example, the case where the contract data is stored in the storage unit 110 in the driver characteristic estimation device 100 has been described, but the present invention is not limited to this example. For example, contract data may be stored in the cloud. In this case, driver characteristic estimation device 100 acquires the contract data of the user ID from the cloud.
In the embodiment described above, as an example, the case where the driver characteristic estimation device 100 associates the vehicle data notification with the acquisition time and stores them in the storage unit 110 has been described, but the present invention is not limited to this example. For example, vehicle data notification and acquisition time may be associated and stored in the cloud. In this case, driver characteristic estimation device 100 acquires information that associates the vehicle data notification with the acquisition time from the cloud.
In the above-described embodiment, as an example, the driver characteristic estimation system 1 includes one portable terminal 200, one vehicle-mounted device 300, and one driver characteristic estimation device 100. However, it is not limited to this example. For example, the driver characteristic estimation system 1 may include a plurality of portable terminals 200, a plurality of vehicle-mounted devices, or a plurality of driver characteristic estimation devices 100. FIG.
In the above-described embodiment, the case where the driver characteristic estimation device 100 estimates one of the Big 5, the driving attitude (DSQ), and the burden sensitivity (WSQ) as an example of the psychological characteristic has been described, but the present invention is not limited to this example. . For example, the driver characteristic estimation device 100 may estimate the age of the driver, or may estimate whether the target driver is driving by identifying the driver.

Here, a method for creating the learned model 109 will be described.
FIG. 9 is a diagram showing an example of the learning device of this embodiment.
(Learning device 400)
The learning device 400 is implemented by a device such as a personal computer, server, smart phone, tablet computer, or industrial computer. The learning device 400 includes an input unit 402, an acquisition unit 403, a collection unit 405, a learning unit 408, a verification unit 409, a storage unit 410, and a selection unit 411, for example.

The input unit 402 receives contract data-related information that associates the contract data of the mobile terminal 200 held by the driver of the vehicle VE with the information specifying the driver's mental state. The input unit 402 obtains vehicle data including sensor information (operation information of the vehicle VE and state information of the vehicle VE) measured by sensors capable of measuring the state of the vehicle VE driven by the driver of the portable terminal 200, and the vehicle data. Vehicle data-related information associated with information specifying the driver's state of mind at the time of driving is input.
The acquisition unit 403 acquires the contract data-related information input to the input unit 402, and causes the storage unit 410 to store the acquired contract data-related information in association with the acquisition time.
The collection unit 405 collects the vehicle data-related information input to the input unit 402, and causes the storage unit 410 to store the collected vehicle data-related information in association with the acquisition time.
The storage unit 410 is implemented by a HDD (Hard Disk Drive), flash memory, RAM (Random Access Memory), ROM (Read Only Memory), or the like. The storage unit 410 stores the contract data related information and the acquisition time in association with each other. The storage unit 410 associates and stores the vehicle data related information and the acquisition time.

The learning unit 408 comprises a first learning model 408-1 and a second learning model 408-2. The learning unit 408 acquires the contract data-related information acquired by the acquisition unit 403, and based on the information that associates the contract data included in the acquired contract data-related information with the information that specifies the psychological state of the driver, the input A first learning model 408-1 is created by performing machine learning using teacher data, which is a set of contract data and information specifying the psychological state of the driver corresponding to the contract data.
The learning unit 408 acquires the vehicle data-related information collected by the collection unit 405, and based on the information that associates the vehicle data included in the acquired vehicle data-related information with the information that specifies the psychological state of the driver, the input is performed. A second learning model 408-2 is created by performing machine learning using teacher data, which is a set of vehicle data and information specifying the psychological state of the driver corresponding to the vehicle data.
For example, the learning unit 408 creates a first learning model 408-1 and a second learning model 408-2 by performing machine learning using an estimation algorithm such as XGBoost (eXtreme Gradient Boosting). XGBoost is a decision tree-based algorithm, also called gradient boosting. XGBoost builds weak learners sequentially. XGBoost has very good generalization performance for various tasks. Learning unit 408 may create first learning model 408-1 and second learning model 408-2 using other algorithms such as random forest, deep learning, and SVM (Support Vector Machine).

After first learning model 408-1 and second learning model 408-2 are created, verification unit 409 inputs contract data into first learning model 408-1 for each index included in the driver's state of mind. and the output (score (hereinafter referred to as "second score")) obtained by inputting the vehicle data to the second learning model 408-2. and get.
The verification unit 409 determines whether the score is high or low based on the acquired first score and second score. Specifically, verification section 409 derives an average value and a standard deviation for each of the first score and the second score. Verification unit 409 determines that the score is high if each of the first score and the second score is equal to or greater than the result of adding the standard deviation to the average value, and determines that the score is high if the result of subtracting the standard deviation from the average value is equal to or less than In some cases, it is determined that the score is low. When the verification unit 409 determines that the score is high, it converts it to "1", and when it determines that the score is not high, it converts it to "0", thereby converting it into a binary value.
The verification unit 409 derives the accuracy of the first score and the accuracy of the second score.

FIG. 10 is a diagram showing an example of processing of the learning device of this embodiment. FIG. 10 shows an example of the verification result by the verification unit 409 when Big5 is used. According to FIG. 10, for each of intellectual curiosity, conscientiousness, extroversion, cooperativeness, and emotional instability, contract data is first learned for high and low scores. The accuracy of the output when input to model 408-1 and the accuracy of the output when vehicle data is input to the second learning model 408-2 are shown.
The selection unit 411 selects the contract data from the first learning model 408 when the scores of intellectual curiosity, conscientiousness, extroversion, agreeableness, and emotional instability are high and low. -1 and the output accuracy when the vehicle data is input to the second learning model 408-2, the one with the highest accuracy is selected. For example, when the intellectual curiosity is high, the accuracy of the output when the contract data is input to the first learning model 408-1, the vehicle data is input to the first learning model 408-1 The vehicle data is selected because the accuracy of the output of is higher.
When the intellectual curiosity is low (low), the accuracy of the output when the contract data is input to the first learning model 408-1 is higher than that when the vehicle data is input to the first learning model 408-1. Select the contract data because it is more accurate than the output of . Use data is selected in a similar manner for other personality indicators.
As a result, the relationship between the index included in the psychological characteristics shown in FIG. 7 and the usage data used when deriving the index is created for each tendency. High scores correspond to high propensity and low scores correspond to low propensity.

First learning model 408 - 1 and second learning model 408 - 2 created by learning unit 408 are input to derivation unit 108 of driver characteristic estimation device 100 . Derivation unit 108 uses input first learning model 408 - 1 and second learning model 408 - 2 as trained model 109 . Information that associates the derivation target selected by selection unit 411 , the tendency of the derivation target, and the usage data used when deriving the tendency of the derivation target is input to derivation unit 108 of driver characteristic estimation device 100 . The derivation unit 108 acquires information that associates a derivation target, a tendency of the derivation target, and use data used when deriving the tendency of the derivation target, and uses the acquired information when deriving the psychological characteristic. do.
By inputting first learning model 408-1 and second learning model 408-2 created by learning device 400 to derivation unit 108 of driver characteristic estimation device 100, learned model 109 of derivation unit 108 is updated. may

According to the driver characteristics estimation system 1 of the present embodiment, the driver characteristics estimation device 100 includes a collection unit 105 that collects vehicle data related to the vehicle VE driven by the driver, and a contract indicating contract details based on the portable terminal owned by the driver. Based on an acquisition unit 103 that acquires data, a learned model 109 that includes the results of machine learning of the relationship between the contract data and the psychological characteristics of the driver, and the results of machine learning of the relationship between the vehicle data and the psychological characteristics of the driver. and a derivation unit 108 for deriving psychological characteristics related to the acquired contract data or vehicle data.
With this configuration, the driver characteristic estimation device 100 includes the results of machine learning of the relationship between the contract data and the psychological characteristics of the driver and the results of machine learning of the relationship between the vehicle data and the psychological characteristics of the driver. Based on the trained model 109, psychological characteristics associated with contract data or vehicle data can be derived. Therefore, in addition to temporarily obtained elements such as vehicle data of the vehicle VE, the driver's mental state can be derived from permanent data such as contract data. can.
In the driver characteristic estimation device 100, the psychological characteristic includes any one of Big 5, driving attitude, and burden sensitivity.
With this configuration, the driver characteristic estimation device 100 includes the results of machine learning of the relationship between the contract data and the psychological characteristics of the driver and the results of machine learning of the relationship between the vehicle data and the psychological characteristics of the driver. Based on the trained model 109, any one of Big 5, driving attitude, and burden sensitivity can be derived as psychological characteristics related to contract data or vehicle data. For this reason, in addition to temporarily obtained elements such as vehicle data of the vehicle VE, permanent data such as contract data can be used to derive any of the driver's Big 5, driving attitude, and burden sensitivity. It is possible to improve the determination accuracy of the mental state of the person.

(Modification of embodiment)
FIG. 1 can be applied to the configuration example of the driver characteristic estimation system 1a of the modified example of the embodiment. However, it differs in that a driver characteristic estimation device 100a is provided instead of the driver characteristic estimation device 100. FIG.
FIG. 11 is a diagram showing an example of processing of the driver characteristic estimation system according to the modified example of the embodiment.
The driver characteristic estimation device 100a associates and registers the user ID of the mobile terminal 200 and the in-vehicle device ID. Here, a vehicle ID may be registered instead of the vehicle-mounted device ID. As an example, the case where the vehicle-mounted device ID is registered will be described below. The driver characteristic estimation device 100a registers the contract data of the user ID. Here, the contract data of the user ID may be registered in a server different from the driver characteristic estimation device 100a. As an example, the case where the contract data of the user ID is registered in the driver characteristic estimation device 100a will be described below.
The in-vehicle device 300a detects operation information of the driver (user) of the vehicle VE and state information of the vehicle VE. The in-vehicle device 300a creates a vehicle data notification including the vehicle data of the vehicle VE including the detected operation information of the vehicle VE and the state information of the vehicle VE and the in-vehicle device ID, and sends the created vehicle data notification to the driver. Transmit to characteristic estimation device 100a (1).

The driver characteristic estimation device 100a receives the vehicle data notification transmitted by the vehicle-mounted device 300a, and acquires the vehicle data included in the received vehicle data notification. The driver characteristic estimation device 100a acquires the user ID linked to the vehicle-mounted device ID based on the acquired vehicle data, and acquires the contract data of the acquired user ID. The driver characteristic estimation device 100a generates a first characteristic amount that quantitatively represents the characteristics of the acquired contract data.
Based on the acquired vehicle data, the driver characteristic estimation device 100a generates a second feature quantity that quantitatively represents the characteristics of the vehicle data (2).
The driver characteristic estimating device 100a is based on a learned model that includes the result of machine learning of the relationship between the first feature amount and the driver's psychological characteristic and the result of machine learning of the relationship between the second feature amount and the driver's psychological characteristic. Then, a psychological characteristic related to the generated first feature amount or the calculated second feature amount is derived (3).
The driver characteristic estimation device 100a creates a psychological characteristic notification including the psychological characteristic derivation result, and outputs the created psychological characteristic notification (4). An example of the psychological characteristic derivation result is any of BIG5, driving attitude (DSQ), and burden sensitivity (WSQ).
Among the driver characteristics estimation device 100a, the portable terminal 200, and the vehicle-mounted device 300a included in the driver characteristics estimation system 1, the driver characteristics estimation device 100a different from the embodiment will be described below.
FIG. 12 is a diagram showing details of a driver characteristic estimation device included in the driver characteristic estimation system of the modified example of the embodiment.

(Driver characteristic estimation device 100a)
Driver characteristic estimation device 100a is realized by a device such as a personal computer, a server, a smart phone, a tablet computer, or an industrial computer. Driver characteristic estimation device 100a includes, for example, communication unit 102, acquisition unit 103, creation unit 104, collection unit 105, first feature amount generation unit 106-1, second feature amount generation unit 106-2, derivation unit 108a, and storage unit. 110.
The first feature amount generation unit 106-1 acquires the contract data and the user ID acquired by the acquisition unit 103, and generates a first feature amount that quantitatively represents the features of the contract data based on the acquired contract data. do.
A second feature quantity generation unit 106-2 acquires the vehicle data collected by the collection unit 105 and the on-vehicle device ID, and generates a second feature quantity that quantitatively expresses the characteristics of the vehicle data based on the collected vehicle data. Generate.

FIG. 13 is a diagram illustrating an example of processing of the driver characteristic estimation device according to the modified example of the embodiment;
Second feature quantity generation unit 106-2 divides the vehicle data into driving habits and driving operations. The second feature quantity generation unit 106-2 generates a driving habit feature quantity that quantitatively represents the characteristics of the driving habit and a driving operation feature quantity that quantitatively represents the characteristics of the driving operation.
A case of generating a driving habit feature amount will be specifically described. The second feature amount generation unit 106-2 generates a feature amount of the time of each driving, a feature amount of the day of the week, a feature amount of the driving distance, and the like from the vehicle data corresponding to the driving habit. The second feature amount generation unit 106-2 may generate an all-day feature amount, a weekday feature amount, a weekend feature amount, and a holiday feature amount from vehicle data corresponding to driving habits. From the data corresponding to driving habits, feature amount for all hours, feature amount for 5:00 to 11:00, feature amount for 11:00 to 17:00, feature amount for 17:00 to 23:00, feature amount for 23:00 to 5:00 quantity can be generated.

The second feature quantity generation unit 106-2 generates a feature quantity of the number of times of driving per day or one trip, a feature quantity of the travel distance, and a feature quantity of the travel time from the vehicle data corresponding to the driving habits. good too.
Second feature amount generation section 106-2 may calculate statistical values, autocorrelation, travel time ratio, travel distance, etc. for the generated feature amounts. A ratio may be calculated.
For example, when calculating the autocorrelation, the second feature value generating section 106-2 may calculate the autocorrelation for a predetermined period such as one day or one week. When calculating the travel time ratio, the second feature amount generation unit 106-2 may calculate the travel time ratio for a predetermined time such as 10 min to 15 min or 15 min to 30 min. When calculating the travel distance ratio, the second feature amount generation unit 106-2 may calculate the travel distance ratio of a predetermined distance such as 0 km to 5 km or 5 km to 10 km.
FIG. 14 is a diagram showing an example of processing of the driver characteristic estimation device according to the modified example of the embodiment. FIG. 14 shows an example of processing for generating a driving habit feature amount from vehicle data. An example of vehicle data is stored in association with acquisition time. In this example, vehicle data corresponding to driving habits are collected every second. The second feature quantity generation unit 106-2 generates a feature quantity for each predetermined time window. In the example shown in FIG. 14, the second feature amount generation unit 106-2 generates a feature amount of driving time, a feature amount of driving distance, and a feature amount indicating whether it is a weekday or a holiday for each trip.

A specific description will be given of the case where the driving operation feature amount is generated. Second feature value generation unit 106-2 divides vehicle data corresponding to driving operations into one or more driving scenes. Examples of driving scenes include a right turn scene, a left turn scene, a parking scene, a scene where the vehicle speed is 5 km/h to 20 km/h, a scene where the vehicle speed is 20 km/h to 40 km/h, and a scene where the vehicle speed is 40 km/h to 60 km/h. is 60 km/h or more.
The second feature amount generation unit 106-2 extracts, for example, a section of 100 m before and after the point where the steering angle exceeds 120 degrees as a right turn scene. The second feature value generation unit 106-2 extracts, for example, a section of 100 m before and after the point where the steering angle of the steering wheel falls below -120 degrees as a left turn scene. Second feature value generation unit 106-2 extracts, as a parking scene, a section within a travel distance of 100 m and including the end of the trip from the point where the shift position is reversed, for example.

The second feature value generation unit 106-2 extracts, for example, a section in which the vehicle speed is between 5 km/h and 20 km/h and continues for 30 seconds or longer as a driving scene where the vehicle speed is between 5 km/h and 20 km/h. The second feature value generation unit 106-2 detects, for example, a section in which the vehicle speed is between 20 km/h and 40 km/h and continues for 60 seconds or more. A driving scene in which the vehicle speed is 20 km/h to 40 km/h is extracted.
The second feature value generator 106-2 extracts, for example, a section in which the vehicle speed is between 40 km/h and 60 km/h and continues for 30 seconds or longer as a driving scene where the vehicle speed is between 40 km/h and 60 km/h. The second feature value generation unit 106-2 extracts, for example, a section in which the vehicle speed is 60 km/h or more and continues for 30 seconds or more as a driving scene of the section in which the vehicle speed is 60 km/h or more.
A second feature amount generation unit 106-2 generates a driving operation feature amount that quantitatively represents the characteristics of the driving operation from the vehicle data corresponding to the driving operation included in each of the extracted one or more driving scenes. . The second feature value generation unit 106-2 generates a right-turn scene, a left-turn scene, a parking scene, a vehicle speed of 5 km/h to 20 km/h from vehicle data corresponding to driving operations included in each of one or a plurality of driving scenes. For each driving scene, such as a scene, a scene where the vehicle speed is 20 km/h to 40 km/h, a scene where the vehicle speed is 40 km/h to 60 km/h, and a scene where the vehicle speed is 60 km/h or more, a fixed time interval such as 3 seconds Generate features.
The second feature value generation unit 106-2 uses statistics of each item obtained by CAN (Controller Area Network) communication from vehicle data corresponding to driving operations included in each of one or more driving scenes to generate features. quantity can be generated. Examples of statistics are maximum, minimum, mean, variance, median, standard deviation, first quartile, third quartile, skewness, kurtosis, and the like. For example, the second feature value generation unit 106-2 uses the vehicle speed statistic, the accelerator depression statistic, and the acceleration statistic for each of the right-turn and left-turn driving scenes to generate feature values. to generate

The second feature value generation unit 106-2 uses FFT (Fast Fourier Transformation) of each item obtained by CAN communication from the vehicle data corresponding to the driving operation included in each of one or more driving scenes to generate the feature value. may be generated. An example of FFT is to decompose into frequency components such as 0 Hz, 1 Hz, 2 Hz, .
The second feature amount generation unit 106-2 generates feature amounts using autocorrelation of each item obtained by CAN communication from vehicle data corresponding to driving operations included in each of one or more driving scenes. good too. An example autocorrelation is 0.2 seconds, 1 second, 5 seconds, 10 seconds, 20 seconds, 30 seconds, 60 seconds, and so on.
FIG. 15 is a diagram showing an example of processing of the driver characteristic estimation device according to the modified example of the embodiment. FIG. 15 shows an example of processing for generating a driving operation feature quantity from vehicle data. An example of vehicle data is stored in association with acquisition time. In this example, vehicle data for each driving scene is collected every second. The second feature quantity generation unit 106-2 generates a feature quantity for each predetermined time window. In the example shown in FIG. 15, the second feature quantity generation unit 106-2 generates a feature quantity of the vehicle speed average, a feature quantity of the vehicle speed standard deviation, and a feature quantity of the maximum (value) of the vehicle speed for each window. Returning to FIG. 12, the description continues.

Derivation unit 108a includes trained model 109a. Derivation unit 108a acquires the first feature amount and the user ID from first feature amount generation unit 106-1, and acquires the second feature amount and the vehicle-mounted device ID from second feature amount generation unit 106-2. The derivation unit 108a derives a psychological characteristic related to the acquired first feature amount or a psychological characteristic related to the second feature amount based on the learned model 109a.
The trained model 109a will be explained. The trained model 109a includes the results of machine learning of the relationship between the first feature quantity and the driver's psychological characteristics, and the results of machine learning of the relationship between the second feature quantity and the driver's psychological characteristics. An example of psychological characteristics is one of Big 5, Driving Attitude (DSQ) and Stress Sensitivity (WSQ). The Big 5 can estimate the driver's personality, the DSQ can estimate the driver's driving psychology, and the WSQ can estimate the driver's driving characteristics.

Whether the learned model 109a is derived using the first feature amount or the second feature amount is set for each tendency of the psychological characteristic index. For example, the trained model 109a stores the relationship between an index (derivation target) included in psychological characteristics, the tendency of the index, and the usage data used when deriving the tendency of the index (derivation target). Here, as an example, the relationship between the five indices included in Big5, the tendency of each of the five indices, and the usage data used when deriving each of the five indices will be described.
Specifically, regarding intellectual curiosity, when deriving a high tendency (high), the feature amount (second feature amount) of the vehicle data is used, and when deriving a low tendency (low), the feature amount (second feature amount) of the contract data is used. 1st feature quantity) is used. Concerning conscientiousness, it is shown that the feature amount of the contract data is used when deriving a high tendency, and the feature amount of the vehicle data is used when deriving a low tendency. Regarding extroversion, it is shown that the feature amount of the contract data is used when deriving a high tendency, and the feature amount of the vehicle data is used when deriving a low tendency (low). Regarding cooperativeness, it is shown that the feature amount of vehicle data is used when deriving a high tendency, and the feature amount of contract data is used when deriving a low tendency (low). Regarding emotional instability, it is shown that the feature amount of the vehicle data is used when deriving a high tendency, and the feature amount of the vehicle data is used when deriving a low tendency (low). Returning to FIG. 12, the description is continued.

The derivation unit 108a inputs the second feature amount to the learned model 109a when deriving a high tendency for intellectual curiosity, and acquires information indicating a high tendency for intellectual curiosity output from the learned model 109a. The derivation unit 108a inputs the first feature amount to the learned model 109a when deriving a tendency of low intellectual curiosity, and acquires information indicating a tendency of low intellectual curiosity output from the trained model 109a. The derivation unit 108a inputs the first feature amount to the learned model 109a when deriving a high sincerity tendency, and acquires information indicating a high sincerity tendency output by the trained model 109a. The derivation unit 108a inputs the second feature amount to the learned model 109a when deriving a tendency of low conscientiousness, and acquires information indicating a tendency of low conscientiousness output by the learned model 109a.
The derivation unit 108a inputs the first feature amount to the learned model 109a when deriving a high tendency for extraversion, and acquires information indicating a high tendency for extraversion output from the trained model 109a. The derivation unit 108a inputs the second feature amount to the learned model 109a when deriving the tendency of low extroversion, and acquires the information indicating the tendency of low extroversion output from the trained model 109a. The derivation unit 108a inputs the second feature amount to the learned model 109a when deriving a tendency of high cooperativeness, and acquires information indicating a tendency of high cooperativeness output from the learned model 109a. The derivation unit 108a inputs the first feature amount to the learned model 109a when deriving a tendency of low cooperativeness, and acquires information indicating a tendency of low cooperativeness output from the learned model 109a.
Derivation unit 108a inputs the second feature amount to learned model 109a when deriving a high tendency for emotional instability, and acquires information indicating a high tendency for emotional instability output by trained model 109a. The derivation unit 108a inputs the second feature amount to the learned model 109a when deriving a low tendency for emotional instability, and acquires information indicating a low tendency for emotional instability output by the trained model 109a.

In the first feature amount generation unit 106-1, the second feature amount generation unit 106-2, and the derivation unit 108a, for example, a hardware processor such as a CPU executes a computer program (software) stored in the storage unit 110. It is realized by
In addition, some or all of these functional units may be realized by hardware (including circuitry) such as LSI, ASIC, FPGA, GPU, etc., or by cooperation of software and hardware. may be implemented. Returning to FIG. 12, the description is continued.

(Operation of driver characteristic estimation system 1a)
FIG. 16 is a diagram showing an example of processing of the driver characteristic estimation system according to the modified example of the embodiment.
Steps S1-2 to S4-2 and steps 8-2 to S10-2 can apply steps S1-1 to S4-1 and steps S6-1 to S8-1 in FIG. Description is omitted.
(Step S5-2)
In driver characteristic estimation device 100a, first feature amount generation unit 106-1 acquires the contract data and user ID acquired by acquisition unit 103, and quantitatively expresses the features of the contract data based on the acquired contract data. The first feature amount is generated.

(Step S6-2)
In the driver characteristic estimation device 100a, the collection unit 105 collects the vehicle data notification received by the communication unit 102, acquires the vehicle data and the vehicle-mounted device ID included in the collected vehicle data notification, and stores the acquisition time in the storage unit 110. be stored in association with A second feature quantity generation unit 106-2 acquires the vehicle data collected by the collection unit 105 and the on-vehicle device ID, and generates a second feature quantity that quantitatively expresses the characteristics of the vehicle data based on the acquired vehicle data. Generate.
(Step S7-2)
In driver characteristic estimation device 100a, derivation section 108a acquires a first feature amount from first feature amount generation section 106-1 and acquires a second feature amount from second feature amount generation section 106-2. The derivation unit 108a derives a psychological characteristic related to the acquired first feature amount or a psychological characteristic related to the acquired second feature amount based on the learned model 109a.

Here, a method for creating the trained model 109a will be described.
FIG. 17 is a diagram illustrating an example of a learning device according to a modification of the embodiment;
(Learning device 400a)
The learning device 400a is implemented by a device such as a personal computer, server, smart phone, tablet computer, or industrial computer. The learning device 400a includes, for example, an input unit 402, an acquisition unit 403, a collection unit 405, a first feature amount generation unit 406-1, a second feature amount generation unit 406-2, a learning unit 408a, a verification unit 409a, and a storage unit 410. and a selection unit 411 .
The above-described first feature quantity generation unit 106-1 can be applied to the first feature quantity generation unit 406-1. The first feature amount generation unit 406-1 acquires the contract data-related information acquired by the acquisition unit 403, and based on the contract data included in the acquired contract data-related information, the first feature amount generation unit 406-1 quantitatively expresses the features of the contract data. 1 Generate a feature amount.
The second feature quantity generation unit 406-2 can apply the second feature quantity generation unit 106-2 described above. A second feature quantity generation unit 406-2 acquires the vehicle data related information collected by the collection unit 405, and based on the vehicle data included in the acquired vehicle data related information, the second feature amount generation unit 406-2 quantitatively expresses the vehicle data characteristics. 2 Generate features.
The first feature amount generation unit 406-1 associates the generated first feature amount with information specifying the driver's state of mind. The second feature amount generation unit 406-2 associates the generated second feature amount with information specifying the driver's state of mind.
18 is a diagram illustrating an example of processing of a learning device according to a modification of the embodiment; FIG. The first feature amount generation unit 406-1 acquires contract data as original data, and generates a first feature amount based on the acquired contract data. The second feature amount generation unit 406-2 acquires vehicle data as original data, and generates a second feature amount based on the acquired vehicle data. The second feature amount includes a driving operation feature amount that quantitatively expresses driving operation features such as a right turn scene, a left turn scene, and a 0 to 20 km/h scene, and a driving habit feature that quantitatively represents the characteristics of driving habits. quantity and. Returning to FIG. 17, the description is continued.

The learning unit 408a comprises a first learning model 408a-1 and a second learning model 408a-2. The learning unit 408a identifies the first feature amount to be input and the psychological state of the driver corresponding to the first feature amount based on the information that associates the first feature amount with the information specifying the driver's mental state. A first learning model 408a-1 is created by performing machine learning using teacher data, which is a set of information to be used.
The learning unit 408a identifies the second feature amount to be input and the psychological state of the driver corresponding to the second feature amount based on the information that associates the second feature amount with the information specifying the driver's state of mind. A second learning model 408a-2 is created by performing machine learning using teacher data, which is a set of information to be used.
For example, the learning unit 408a creates a first learning model 408a-1 and a second learning model 408a-2 by performing machine learning using an estimation algorithm such as XGBoost. The learning unit 408a may create the first learning model 408a-1 and the second learning model 408a-2 using other algorithms such as random forest, deep learning, and SVM.

After the first learning model 408a-1 and the second learning model 408a-2 are created, the verification unit 409a applies the first feature amount to the first learning model 408a-1 for each index included in the driver's state of mind. and a second score obtained by inputting the second feature amount to the second learning model 408a-2.
The verification unit 409a determines whether the score is high or low based on the acquired first score and second score. Specifically, the verification unit 409a derives an average value and a standard deviation for each of the first score and the second score. The verification unit 409a determines that the score is high when each of the first score and the second score is equal to or greater than the result of adding the standard deviation to the average value, and determines that the score is high when the result of subtracting the standard deviation from the average value is equal to or less than the result of subtracting the standard deviation from the average value. In some cases, it is determined that the score is low. When the verification unit 409a determines that the score is high, it converts it to "1", and when it determines that the score is not high, it converts it to "0", thereby converting it into a binary value.
The verification unit 409a derives the accuracy of the first score and the accuracy of the second score.

19 is a diagram illustrating an example of processing of a learning device according to a modification of the embodiment; FIG. FIG. 19 shows an example of the verification result by the verification unit 409a when Big5 is used. According to FIG. 19, the first feature amount (service data) is input to the first learning model 408a-1, and the second feature amount (driving habit, left turn, right turn, speed 5-20 km/h, speed 20-40 km/h, speed 40-60 km/h, speed 60 km/h -, parking) are input to the second learning model 408a-2.

First learning model 408a-1 and second learning model 408a-2 created by learning unit 408a are input to derivation unit 108a of driver characteristic estimation device 100a. Derivation unit 108a uses input first learning model 408a-1 and second learning model 408a-2 as trained model 109a. Information that associates the tendency of the derivation target and the detection target selected by the selection unit 411 with the feature amount of the used data used when deriving the tendency of the derivation target is input to the derivation unit 108a of the driver characteristic estimation device 100a. be. The derivation unit 108a obtains information that associates a derivation target, a tendency of the derivation target, and a feature amount of the use data used when deriving the tendency of the derivation target, and derives a psychological characteristic from the obtained information. Use sometimes.
By inputting the first learning model 408a-1 and the second learning model 408a-2 created by the learning device 400a to the derivation unit 108a of the driver characteristic estimation device 100a, the learned model 109a of the derivation unit 108a is updated. You may

According to the driver characteristic estimation system 1a of the modified example of the embodiment, the driver characteristic estimation device 100a includes the collection unit 105 that collects vehicle data related to the vehicle VE driven by the driver, and the content of the contract based on the portable terminal possessed by the driver. Acquisition unit 103 for acquiring contract data indicated by acquisition unit 103; a second feature quantity generation unit 106-2 for generating a second feature quantity that quantitatively expresses the characteristics of the vehicle data of the vehicle VE, and the results of machine learning of the relationship between the first feature quantity and the psychological characteristics of the driver; a derivation unit 108a for deriving a psychological characteristic related to the first feature amount or the second characteristic amount based on a trained model 109a including the result of machine learning of the relationship between the second characteristic amount and the driver's psychological characteristic; Prepare.
With this configuration, the driver characteristic estimating device 100a generates the first characteristic quantity that quantitatively represents the characteristic of the contract data, and the second characteristic quantity that quantitatively represents the characteristic of the vehicle data of the vehicle VE. based on the trained model 109a including the result of machine learning of the relationship between the first feature quantity and the driver's psychological characteristics and the result of machine learning of the relationship between the second feature quantity and the driver's psychological characteristics. , a psychological characteristic associated with the first feature or the second feature. Therefore, the driver's psychological state can be derived from permanent data such as contract data in addition to temporarily obtained elements such as vehicle data of the vehicle VE. can.

In the driver characteristic estimation device 100a, the second feature value generation unit 106-2 divides the vehicle data of the vehicle VE into the driving habits of the driver of the vehicle VE and the driving operations of the driver of the vehicle, and determines the driving habits of the driver of the vehicle VE. and a driving operation feature amount quantitatively representing the characteristics of the driving operation of the driver of the vehicle VE.
With such a configuration, it is possible to derive the psychological characteristics related to the driving habit feature quantity of the driver of the vehicle VE and the psychological characteristics related to the driving operation characteristic quantity of the driver of the vehicle VE, so that the determination accuracy of the psychological state of the driver can be obtained. can be improved.
In the driver characteristic estimation device 100a, the second feature value generating unit 106-2 divides the driving operation of the driver into a plurality of driving scenes, and based on the driving operation of the vehicle driver included in each of the plurality of driving scenes, 2 Generate features.
By configuring in this manner, it is possible to derive the psychological characteristics related to the second feature amount generated based on the driver's driving operation in each of the plurality of driving situations, thereby improving the determination accuracy of the driver's psychological state. be able to.

In the driver characteristic estimation device 100a, the derivation unit 108a obtains the result of machine learning of the relationship between the first feature amount and the driver's psychological characteristic according to a plurality of derived items included in the psychological characteristic, and the second characteristic amount and the driver's psychological characteristics related to the first feature quantity or the psychological characteristics related to the second feature quantity are derived based on either the results of machine learning of the relationship with the psychological characteristics of .
By configuring in this way, according to a plurality of derived items included in the psychological characteristics, the result of machine learning of the relationship between the first feature quantity and the psychological characteristics of the driver, the second characteristic quantity and the psychological characteristics of the driver. Since the learned model 109a including the result of machine learning of the relationship between the two can be used properly, it is possible to improve the determination accuracy of the driver's state of mind.

In the driver characteristic estimation device 100a, for each of the plurality of derived items included in the psychological characteristic, the accuracy of the derived item obtained from the learned model and the first feature amount, and the accuracy of the derived item obtained from the learned model and the second feature amount Either the first feature amount or the second feature amount is used when deriving the derived item based on the accuracy of the derived item.
With this configuration, for each of the plurality of derived items included in the psychological characteristic, the accuracy of the derived item obtained from the learned model and the first feature amount, and the accuracy of the derived item obtained from the learned model and the second feature amount. Since the first feature amount and the second feature amount can be selectively used based on the accuracy of the derived items obtained, the determination accuracy of the driver's state of mind can be improved.

Although the embodiments of the present invention have been described in detail above with reference to the drawings, the specific configuration is not limited to these embodiments, and design changes and the like are also included within the scope of the present invention. For example, a computer program for realizing the functions of the devices described above may be recorded on a computer-readable recording medium, and the computer program recorded on this recording medium may be read and executed by a computer system. . Note that the “computer system” referred to here may include hardware such as an OS and peripheral devices.
In addition, "computer-readable recording medium" includes writable nonvolatile memories such as flexible discs, magneto-optical discs, ROMs and flash memories, portable media such as DVDs (Digital Versatile Discs), and computer system built-in media. A storage device such as a hard disk that Furthermore, "computer-readable recording medium" means a volatile memory (e.g., DRAM ( It also includes those that hold programs for a certain period of time, such as Dynamic Random Access Memory)).
Further, the above program may be transmitted from a computer system storing this program in a storage device or the like to another computer system via a transmission medium or by a transmission wave in a transmission medium. Here, the "transmission medium" for transmitting the program refers to a medium having a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line.
Further, the program may be for realizing part of the functions described above.
Further, it may be a so-called difference file (difference program) that can realize the above-described functions in combination with a program already recorded in the computer system.

Reference Signs List 1, 1a... Driver characteristic estimation system 100, 100a... Driver characteristic estimation device 102... Communication unit 103... Acquisition unit 104... Creation unit 105... Collection unit 106-1... First feature quantity generation unit 106 -2... Second feature quantity generation unit 108, 108a... Derivation unit 109, 109a... Learned model 110... Storage unit 200... Portable terminal 202... Communication unit 209... Output unit 210... Storage unit, 300... vehicle-mounted device, 302... communication unit, 304, 304a... creation unit, 306... detection unit, 307... positioning unit, 309... output unit, 310... storage unit, 400... learning device, 402... input unit, 403... acquisition Unit 405... Collection unit 406-1... First feature amount generation unit 406-2... Second feature amount generation unit 408, 408a... Learning unit 408-1, 408a-1... First learning model 408 -2, 408a-2... second learning model, 409, 409a... verification unit, 410... storage unit, 411... selection unit

(1) One aspect of the present invention includes a collection unit that collects vehicle data related to a vehicle driven by a driver, an acquisition unit that acquires contract data indicating contract details based on a mobile terminal owned by the driver, and contract data. Relationship between the acquired contract data or the vehicle data based on a learned model including the result of machine learning of the relationship with the psychological characteristics of the driver and the result of machine learning of the relationship between the vehicle data and the psychological characteristics of the driver. a derivation unit for deriving a psychological characteristic to be used, wherein the derivation unit calculates the accuracy of the derived item obtained from the trained model and the contract data for each of a plurality of derived items included in the psychological characteristic, and the learned Based on the accuracy of the derived item obtained from the model and the vehicle data, and based on the information indicating which of the contract data and the vehicle data is used when deriving the derived item. Each of a plurality of derived items included in the psychological characteristics is derived using the contract data or vehicle data obtained from the contract data, and the contract data includes the content of the contract with the mobile phone operator and the information provided by the operator through the mobile terminal It is a driver characteristic estimation system that is contract data indicating contract details with .
(2) In one aspect of the present invention, in the driver characteristic estimation system described in (1) above, a first characteristic quantity that quantitatively represents a characteristic of the contract data acquired by the acquisition unit is generated. a quantity generation unit; and a second feature quantity generation unit for generating a second feature quantity that quantitatively expresses the characteristics of the vehicle data of the vehicle collected by the collection unit, wherein the derivation unit includes a first Based on the learned model including the results of machine learning of the relationship between the feature quantity and the driver's psychological characteristics and the result of machine learning of the relationship between the second feature quantity and the driver's psychological characteristics, the first feature quantity or A psychological characteristic related to the second feature is derived.
(3) In one aspect of the present invention, in the driver characteristic estimation system described in (2) above, the second feature value generation unit converts the vehicle data of the vehicle into the driving habits of the driver of the vehicle and the driving habits of the driver of the vehicle. driving behavior of the driver of the vehicle; a driving habit feature quantity that quantitatively represents the characteristics of the driving habit of the driver of the vehicle; and a driving behavior that quantitatively represents the characteristics of the driving behavior of the driver of the vehicle. Generate features.
(4) In one aspect of the present invention, in the driver characteristic estimation system described in (3) above, the second feature value generation unit divides the driving operation of the driver into a plurality of driving situations, and divides the driving operation into a plurality of driving situations. A second feature amount is generated based on the driving operation of the vehicle driver included in each of the.
(5) In one aspect of the present invention, in the driver characteristic estimation system according to any one of (2) to (4) above, the derivation unit may Then, based on either the result of machine learning of the relationship between the first feature amount and the psychological characteristic of the driver or the result of machine learning of the relationship between the second feature amount and the psychological characteristic of the driver, the first characteristic A psychological characteristic related to the quantity or a psychological characteristic related to the second feature amount is derived.
(6) In one aspect of the present invention, in the driver characteristic estimation system according to any one of (2) to (5) above, for each of the plurality of derived items included in the psychological characteristic, the learned the first feature amount and the second feature amount based on the accuracy of the derived item obtained from the model and the first feature amount and the accuracy of the derived item obtained from the trained model and the second feature amount; is used when deriving derived items.
(7) An aspect of the present invention is the driver characteristic estimation system according to any one of (1) to (6) above, wherein the psychological characteristic is any one of Big 5, driving attitude, and burden sensitivity. or
(8) One aspect of the present invention is the driver characteristic estimation system according to any one of (1) to (7) above, wherein the content of the contract with the mobile phone operator is the color of the terminal, the Via a mobile device, including the manufacturer, model name, OS, model change/purchase date, communication plan fee, data capacity, whether or not you have a communication flat-rate plan contract, or whether or not you use Smart Value (a set discount with a fixed line). The content of the above contract with the business operator made by Presence / Monthly usage fee, Internet bank subscription, new electric power subscription

( 9 ) One aspect of the present invention is a driver characteristic estimation method executed by a computer, comprising: a step of collecting vehicle data relating to a vehicle driven by a driver; Acquisition based on a trained model including a step of acquiring data, the result of machine learning of the relationship between contract data and the driver's psychological characteristics, and the result of machine learning of the relationship between vehicle data and the driver's psychological characteristics and deriving a psychological characteristic related to the contract data or the vehicle data , wherein the deriving step includes, for each of a plurality of derived items included in the psychological characteristic, from the learned model and the contract data. Either the contract data or the vehicle data is used when deriving a derived item, based on the accuracy of the derived item obtained and the accuracy of the derived item obtained from the trained model and the vehicle data. Each of a plurality of derived items included in psychological characteristics is derived using the obtained contract data or vehicle data, and the contract data is the content of the contract with the mobile phone operator. and a method of estimating driver characteristics, which is contract data indicating the details of a contract with a business operator made via a mobile terminal .

( 10 ) One aspect of the present invention includes the steps of: collecting vehicle data related to a vehicle driven by a driver in a computer; acquiring contract data indicating contract details based on a portable terminal owned by the driver; and the driver's psychological characteristics, based on a trained model that includes the results of machine learning of the relationship between the vehicle data and the driver's psychological characteristics, and the obtained contract data or the vehicle data. a step of deriving related psychological characteristics, and in the deriving step, for each of a plurality of derived items included in the psychological characteristics, accuracy of derived items obtained from the learned model and contract data; Based on the accuracy of the derived item obtained from the trained model and the vehicle data, and based on information indicating whether either the contract data or the vehicle data is used when deriving the derived item. , using the acquired contract data or vehicle data to derive each of a plurality of derived items included in the psychological characteristics, the contract data being the content of the contract with the mobile phone operator and the mobile terminal It is a computer program, which is contract data indicating the content of a contract with a business operator .

(1) One aspect of the present invention includes a collection unit that collects vehicle data related to a vehicle driven by a driver, an acquisition unit that acquires contract data indicating contract details based on a mobile terminal owned by the driver, and contract data. The acquired contract data based on a trained model including a first learning model obtained by machine-learning the relationship between the psychological characteristics of the driver and a second learning model obtained by machine-learning the relationship between the vehicle data and the psychological characteristics of the driver. or a derivation unit for deriving psychological characteristics related to the vehicle data, wherein the derivation unit obtains each of a plurality of derived items included in the psychological characteristics from the first learning model and contract data Used when either the contract data or the vehicle data derives a derived item based on the accuracy of the derived item and the accuracy of the derived item obtained from the second learning model and the vehicle data. Each of a plurality of derived items included in the psychological characteristic is derived using the acquired contract data or vehicle data based on the information representing whether the It is a driver characteristic estimation system, which is contract data indicating the contents of a contract with a business operator made via a mobile terminal.
(2) In one aspect of the present invention, in the driver characteristic estimation system described in (1) above, a first characteristic quantity that quantitatively represents a characteristic of the contract data acquired by the acquisition unit is generated. a quantity generation unit; and a second feature quantity generation unit for generating a second feature quantity that quantitatively expresses the characteristics of the vehicle data of the vehicle collected by the collection unit, wherein the derivation unit includes a first Based on a trained model including a first learning model that machine-learned the relationship between the feature quantity and the driver's psychological characteristics, and a second learning model that machine-learned the relationship between the second feature quantity and the driver's psychological characteristics, A psychological characteristic related to the first feature amount or the second feature amount is derived.
(3) In one aspect of the present invention, in the driver characteristic estimation system described in (2) above, the second feature value generation unit converts the vehicle data of the vehicle into the driving habits of the driver of the vehicle and the driving habits of the driver of the vehicle. driving behavior of the driver of the vehicle; a driving habit feature quantity that quantitatively represents the characteristics of the driving habit of the driver of the vehicle; and a driving behavior that quantitatively represents the characteristics of the driving behavior of the driver of the vehicle. Generate features.
(4) In one aspect of the present invention, in the driver characteristic estimation system described in (3) above, the second feature value generation unit divides the driving operation of the driver into a plurality of driving situations, and divides the driving operation into a plurality of driving situations. A second feature amount is generated based on the driving operation of the vehicle driver included in each of the.
(5) In one aspect of the present invention, in the driver characteristic estimation system according to any one of (2) to (4) above, the derivation unit may Based on either a first learning model obtained by machine-learning the relationship between the first feature quantity and the driver's psychological characteristics, or a second learning model obtained by machine-learning the relationship between the second feature quantity and the driver's psychological characteristics. Then, a psychological characteristic related to the first feature amount or a psychological characteristic related to the second feature amount is derived.
(6) In one aspect of the present invention, in the driver characteristic estimation system according to any one of (2) to (5) above, for each of a plurality of derived items included in the psychological characteristic, the first Based on the accuracy of the derived item obtained from the learning model and the first feature amount and the accuracy of the derived item obtained from the second learning model and the second feature amount, the first feature amount and Either of the second feature amount is used when deriving a derivation item.
(7) An aspect of the present invention is the driver characteristic estimation system according to any one of (1) to (6) above, wherein the psychological characteristic is any one of Big 5, driving attitude, and burden sensitivity. or
(8) One aspect of the present invention is the driver characteristic estimation system according to any one of (1) to (7) above, wherein the content of the contract with the mobile phone operator is the color of the terminal, the Via a mobile device, including the manufacturer, model name, OS, model change/purchase date, communication plan fee, data capacity, whether or not you have a communication flat-rate plan contract, or whether or not you use Smart Value (a set discount with a fixed line). The content of the above contract with the business operator made by Presence / Monthly usage fee, Internet bank subscription, new electric power subscription

(9) One aspect of the present invention is a driver characteristic estimation method executed by a computer, comprising: a step of collecting vehicle data relating to a vehicle driven by a driver; A trained model including a step of acquiring data, a first learning model machine-learned on the relationship between the contract data and the psychological characteristics of the driver, and a second learning model machine-learned on the relationship between the vehicle data and the psychological characteristics of the driver. and deriving psychological characteristics related to the acquired contract data or vehicle data based on the model, and in the deriving step, for each of the plurality of derived items included in the psychological characteristics, the first Based on the accuracy of the derived items obtained from the learning model and the contract data and the accuracy of the derived items obtained from the second learning model and the vehicle data, either the contract data or the vehicle data is determined. Deriving each of a plurality of derived items included in the psychological characteristics using the acquired contract data or vehicle data based on information indicating whether or not to be used when deriving derived items, The contract data is a driver characteristic estimation method, which is contract data indicating contract details with a mobile phone operator and contract details with the operator made via a mobile terminal.

(10) One aspect of the present invention is a step of collecting vehicle data related to a vehicle driven by a driver in a computer; and the driver's psychological characteristics, and a second learning model that machine-learned the relationship between the vehicle data and the driver's psychological characteristics. deriving a psychological characteristic related to the data or the vehicle data, and in the deriving step, each of a plurality of derived items included in the psychological characteristic is obtained from the first learning model and the contract data. When either the contract data or the vehicle data derives a derived item based on the obtained accuracy of the derived item and the accuracy of the derived item obtained from the second learning model and the vehicle data. Each of a plurality of derived items included in psychological characteristics is derived using the acquired contract data or vehicle data based on the information representing whether the contract data is used for the mobile phone operator and and the contract data indicating the contract content with the business operator made via the mobile terminal.

Claims (9)

a collection unit that collects vehicle data related to a vehicle driven by a driver;
an acquisition unit for acquiring contract data indicating contract details based on a mobile terminal possessed by the driver;
Acquired contract data or vehicle data based on a trained model including the results of machine learning of the relationship between contract data and driver's psychological characteristics and the results of machine learning of the relationship between vehicle data and driver's psychological characteristics and a derivation unit for deriving psychological characteristics related to the driver characteristics estimation system. a first feature quantity generation unit that generates a first feature quantity that quantitatively represents the characteristics of the contract data acquired by the acquisition unit;
a second feature quantity generation unit that generates a second feature quantity quantitatively representing the characteristics of the vehicle data of the vehicle collected by the collection unit;
The derivation unit is based on a trained model including a result of machine learning of the relationship between the first feature amount and the psychological characteristic of the driver and a result of machine learning of the relationship between the second feature amount and the psychological characteristic of the driver. 2. The system for estimating driver characteristics according to claim 1, which derives psychological characteristics related to the first feature quantity or the second feature quantity. The second feature quantity generation unit divides the vehicle data of the vehicle into driving habits of the driver of the vehicle and driving operations of the driver of the vehicle, and quantifies the characteristics of the driving habits of the driver of the vehicle. 3. The driver characteristic estimating system according to claim 2, which generates a characteristic quantity of driving habits expressed in a quantitative manner and a characteristic quantity of driving performance quantitatively expressing characteristics of the driving operation of the driver of the vehicle. The second feature amount generation unit divides the driving operation of the driver into a plurality of driving scenes, and generates a second feature amount based on the driving operation of the vehicle driver included in each of the plurality of driving scenes. The driver characteristic estimation system according to claim 3. The derivation unit performs machine learning of a relationship between a first feature amount and a driver's psychological characteristic and a relationship between a second feature amount and the driver's psychological characteristic according to a plurality of derived items included in the psychological characteristic. Any one of claims 2 to 4, wherein a psychological characteristic related to the first feature amount or a psychological characteristic related to the second feature amount is derived based on one of the results of machine learning. The driver characteristic estimation system described in . Accuracy of a derived item obtained from the learned model and the first feature amount and accuracy of the derived item obtained from the learned model and the second feature amount for each of a plurality of derived items included in the psychological characteristic The driver characteristic according to any one of claims 2 to 5, wherein either the first feature amount or the second feature amount is used when deriving a derivation item based on estimation system. The driver characteristic estimation system according to any one of claims 1 to 6, wherein the psychological characteristic includes any one of Big 5, driving attitude, and burden sensitivity. A computer-implemented driver characteristic estimation method comprising:
collecting vehicle data about the vehicle driven by the driver;
a step of acquiring contract data indicating contract details based on the portable terminal possessed by the driver;
The acquired contract data or the vehicle based on a learned model including the results of machine learning of the relationship between the contract data and the driver's psychological characteristics and the machine learning results of the relationship between the vehicle data and the driver's psychological characteristics. and deriving psychological traits associated with the data. to the computer,
collecting vehicle data about the vehicle driven by the driver;
a step of acquiring contract data indicating contract details based on the portable terminal possessed by the driver;
The acquired contract data or the vehicle based on a learned model including the results of machine learning of the relationship between the contract data and the driver's psychological characteristics and the machine learning results of the relationship between the vehicle data and the driver's psychological characteristics. A computer program causing the steps of deriving psychological properties associated with the data to be performed.

Images