Driving diagnosis device, driving diagnosis system, machine learning device, and method for generating trained model

Abstract

To provide a driving diagnostic device, a driving diagnostic system, a machine learning device, and a learned model generation method considering presence / absence of a passenger in a vehicle.SOLUTION: A driving diagnostic device includes: an acquisition section for acquiring travel data which indicates a travel state of a vehicle; a driving evaluation section for evaluating a driving operation of a driver based on the travel data acquired by the acquisition section; a passenger presence / absence detection section for detecting presence / absence of a passenger in the vehicle; and a notification section for notifying a driver of a driving evaluation result detected by the driving evaluation section and also changing a notification method of the driving evaluation result based on the detection result of the passenger presence / absence detection section.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] The present disclosure relates to a driving diagnosis device, a driving diagnosis system, a machine learning device, and a method for generating a trained model. [Background technology]

[0002] Conventionally, it has been practiced to diagnose whether or not a driving operation of a specific vehicle corresponds to dangerous driving or driving that deteriorates fuel efficiency, and to notify the driver of the vehicle of the results of the driving diagnosis.

[0003] For example, the following Patent Document 1 describes a device that can acquire driving data indicating the vehicle's driving state, evaluate the acquired driving data based on predetermined diagnostic items, and determine the driver's driving tendencies and notify the driver of them, thereby informing the driver of the driver's driving tendencies. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2010-231776 A Summary of the Invention [Problem to be solved by the invention]

[0005] Driving diagnosis as shown in the above Patent Document 1 is expected to have a certain effect in improving the driving quality of drivers. The above-mentioned driving diagnosis can be applied not only to private cars, but also to passenger transport vehicles for carrying passengers (passengers) such as taxis, buses, and vehicles providing ride-sharing services (hereinafter, these are collectively referred to as "commercial vehicles" or simply "vehicles"). If the driving quality of drivers is improved by performing driving diagnosis on commercial vehicles, the quality of passenger transport services can be improved.

[0006] The results of the driving diagnosis are generally reported to the driver one by one through a speaker or a monitor mounted on the vehicle or the like. This reporting method is preferable because the results of the driving diagnosis can be reported to the driver in a timely and reliable manner when there are no passengers in the business vehicle (hereinafter, this state is also referred to as an "empty vehicle"). However, if the results of the driving diagnosis are reported to the driver by the above-mentioned method when there are passengers in the business vehicle (hereinafter, this state is also referred to as an "occupied vehicle"), the results of the driving diagnosis may also be reported to the passengers. Since the information on the results of the driving diagnosis is unnecessary for the passengers, the report may reduce the comfort (ride comfort) of the passengers. In addition, depending on the contents of the results of the driving diagnosis, it may cause anxiety about the driver's driving skills, and may cause passengers to lose trust in the driver.

[0007] In view of the above-mentioned problems, the present disclosure aims to provide a driving diagnosis device, a driving diagnosis system, a machine learning device, and a method for generating a trained model that take into account the presence or absence of passengers in a vehicle. [Means for solving the problem]

[0008] In order to achieve the above object, the driving diagnosis device according to claim 1 of the present disclosure includes an acquisition unit that acquires driving data indicating a driving state of a vehicle, a driving evaluation unit that evaluates the driving operation of a driver of the vehicle based on the driving data acquired by the acquisition unit, a passenger presence / absence detection unit that detects the presence or absence of a passenger in the vehicle, and a notification unit that notifies the driver of the driving evaluation result detected by the driving evaluation unit and changes a notification method of the driving evaluation result based on the detection result of the passenger presence / absence detection unit. The driving evaluation unit evaluates the driving operation of the driver based on the number of times the driving data acquired by the acquisition unit meets the conditions of a plurality of evaluation items related to predetermined driving operation, and evaluates the driving operation of the driver using a first group of evaluation items used when the passenger presence / absence detection unit detects that there is no passenger, and a second group of evaluation items different from the first group of evaluation items used when the passenger presence detection unit detects that there is a passenger.

[0009] In the driving diagnosis device according to claim 1, the method of notifying the driving evaluation result is changed based on the presence or absence of passengers in the vehicle, so that a notification method that takes into account the presence or absence of passengers in the vehicle can be adopted. In addition, when evaluating driving performance, two different groups of evaluation items are used depending on whether or not a passenger is present, making it possible to more accurately evaluate the driver's driving performance.

[0014] Claims of the present disclosure 2The driving diagnosis device according to the present invention includes an acquisition unit that acquires driving data indicating the driving state of a vehicle, a driving evaluation unit that evaluates the driving operation of the driver of the vehicle based on the driving data acquired by the acquisition unit, a passenger presence / absence detection unit that detects the presence or absence of a passenger in the vehicle, and a notification unit that notifies the driver of the driving evaluation result detected by the driving evaluation unit and changes a notification method of the driving evaluation result based on the detection result of the passenger presence / absence detection unit, wherein the passenger presence / absence detection unit includes an inference unit that estimates the presence or absence of a passenger from the driving data, and the inference unit outputs an estimation result of the presence or absence of a passenger by inputting the driving data to an input layer of a trained model that has undergone machine learning to estimate the presence or absence of a passenger.

[0015] Claim 2 In the driving diagnosis device according to the present invention, the notification method of the driving evaluation result is changed based on the presence or absence of a passenger in the vehicle, so that a notification method that takes into account the presence or absence of a passenger in the vehicle can be adopted. In addition, since the presence or absence of a passenger can be estimated by the inference unit, various facilities and installation work for detecting the presence or absence of a passenger are not required.

[0016] Claims of the present disclosure 3 The driving diagnosis system according to the present invention is a driving diagnosis system comprising a server and a terminal device installed in a vehicle, the server and terminal device being communicatively connected to each other via a network, the server and terminal device being installed in the vehicle, the server and terminal device being connected to each other via a network, the server and terminal device being installed in the vehicle, and the system includes an acquisition unit that acquires driving data indicating the driving state of the vehicle, a driving evaluation unit that evaluates the driving operation of the driver of the vehicle based on the driving data acquired by the acquisition unit, a passenger presence / absence detection unit that detects whether or not there is a passenger in the vehicle, and a notification unit that is provided in the terminal device and notifies the driver of the driving evaluation result by the driving evaluation unit and changes the method of notifying the driving evaluation result based on the detection result of the passenger presence / absence detection unit, the driving evaluation unit evaluating the driving operation of the driver based on the number of times that the driving data acquired by the acquisition unit satisfies the conditions of a plurality of evaluation items related to predetermined driving operation, and the system evaluates the driving operation of the driver using a first group of evaluation items used when the passenger presence / absence detection unit detects that there is no passenger, and a second group of evaluation items different from the first evaluation item group that is used when the passenger presence / absence detection unit detects that there is a passenger. Also , Books Disclosed claims 4 The driving diagnosis system according to the present invention is a driving diagnosis system comprising a server and a terminal device installed in a vehicle, the server and terminal devices being communicatively connected to each other via a network, and the system includes an acquisition unit that acquires driving data indicating the driving state of the vehicle, a driving evaluation unit that evaluates the driving operation of the driver of the vehicle based on the driving data acquired by the acquisition unit, a passenger presence / absence detection unit that detects whether or not a passenger is present in the vehicle, and a notification unit that is provided in the terminal device and notifies the driver of the driving evaluation result by the driving evaluation unit and changes a method of notifying the driving evaluation result based on the detection result of the passenger presence / absence detection unit, the passenger presence / absence detection unit including an inference unit that estimates the presence or absence of a passenger from the driving data, and the inference unit outputs an estimation result of the presence or absence of a passenger by inputting the driving data into an input layer of a trained model that has undergone machine learning to estimate the presence or absence of a passenger.

[0017] Claim 3 In the driving diagnosis system according to the present invention, the notification method of the driving evaluation result is changed depending on the presence or absence of passengers in the vehicle, so that a notification method that takes into account the presence or absence of passengers in the vehicle can be adopted. In addition, when evaluating the driving performance, evaluation is performed using two different evaluation item groups depending on the presence or absence of passengers, so that the evaluation of the driver's driving performance can be performed more accurately. Also , request request 4 In the driving diagnosis system according to the present invention, the notification method of the driving evaluation result is changed based on the presence or absence of a passenger in the vehicle, so that a notification method that takes into account the presence or absence of a passenger in the vehicle can be adopted. In addition, since the presence or absence of a passenger can be estimated by the inference unit, various facilities and installation work for detecting the presence or absence of a passenger are not required.

[0018] Claims of the present disclosure 5 The machine learning device according to the present invention is 2The machine learning device for acquiring a learned model used in the inference unit of the driving diagnosis device described in the above, includes: a learning dataset acquisition unit that acquires a plurality of learning datasets each including input data including driving data indicating a driving state of a vehicle and output data including data indicating the presence or absence of a passenger in the vehicle; and a learning processing unit that uses the plurality of learning datasets acquired by the learning dataset acquisition unit to learn a learning model that estimates a correlation between the input data and the output data.

[0019] Claim 5 With the machine learning device according to the present invention, a trained model capable of estimating the presence or absence of a passenger in a vehicle can be obtained from vehicle driving data.

[0020] Claims of the present disclosure 6 The method for generating a trained model according to the present invention comprises the steps of: 2 The method for generating a trained model used in the inference unit of the driving diagnosis device described in the above includes the steps of: acquiring a plurality of training data sets each composed of input data including driving data indicating the driving state of a vehicle and output data including data indicating the presence or absence of a passenger in the vehicle; and using the plurality of training data sets, training a training model that estimates a correlation between the input data and the output data.

[0021] Claim 6 The method for generating a trained model according to the present invention makes it possible to generate a trained model capable of estimating the presence or absence of a passenger in a vehicle from the vehicle's driving data. Effect of the Invention

[0022] According to the present disclosure, it is possible to provide a driving diagnosis device, a driving diagnosis system, a machine learning device, and a method for generating a trained model that take into account the presence or absence of passengers in a vehicle. [Brief description of the drawings]

[0023] [Figure 1] 1 is a schematic diagram showing an overall configuration of an example of a driving diagnosis system according to a first embodiment. [Diagram 2] 1 is a block diagram showing a hardware configuration of an example of a driving diagnosis device according to a first embodiment. [Diagram 3] 1 is a block diagram showing a functional configuration of an example of a driving diagnosis device according to a first embodiment. [Figure 4] 4 is a diagram showing an example of a first evaluation item group of the driving evaluation unit shown in FIG. 3. [Diagram 5] 4 is a diagram showing an example of a second evaluation item group of the driving evaluation unit shown in FIG. 3. [Figure 6] 4 is a flowchart showing an example of a driving diagnosis process by the driving diagnosis device according to the first embodiment of the present disclosure. [Figure 7] FIG. 11 is a block diagram showing a hardware configuration of an example of a driving diagnosis device according to a second embodiment. [Figure 8] FIG. 11 is a block diagram showing a functional configuration of an example of a driving diagnosis device according to a second embodiment. [Figure 9] FIG. 11 is a block diagram showing a functional configuration of an example of a machine learning device according to a second embodiment. [Figure 10] 13 is a flowchart showing an example of a method for generating a trained model according to the second embodiment. [Figure 11] 10 is a flowchart showing an example of a driving diagnosis process performed by a driving diagnosis device according to a second embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0024] Hereinafter, an embodiment for carrying out the present disclosure will be described with reference to the drawings. Note that, in the following, the scope necessary for the explanation to achieve the object of the present disclosure is shown in a schematic manner, and the scope necessary for the explanation of the relevant part of the present disclosure will be mainly explained, and the parts that are omitted from the explanation will be referred to as publicly known techniques.

[0025] <First embodiment> Fig. 1 is a schematic diagram showing an overall configuration of an example of a driving diagnosis system according to a first embodiment of the present disclosure. As shown in Fig. 1, the driving diagnosis system 1 according to the first embodiment includes at least a server 2 and a terminal device 3 installed in a vehicle (business vehicle) V. The server 2 and the terminal device 3 are connected to each other so as to be able to communicate with each other via a network NW. In Fig. 1, one taxi V1 and one ride-sharing service providing vehicle V2 are illustrated as examples of the vehicle V, but the vehicle V is not limited to these and may be another passenger transport vehicle such as a bus.

[0026] The server 2 can be configured as a server computer owned by the manager of the vehicle V. Specifically, the server 2 may be a server owned by a taxi company that manages multiple taxis V1, or a server owned by a company that provides a ride-sharing service included in a terminal device 3 installed in a ride-sharing service providing vehicle V2.

[0027] A well-known computer can be used for this server 2. The well-known computer may include at least a processor, a memory, and a communication interface. This server 2 may be capable of, for example, managing the operation of the vehicle V. In addition, this server 2 may be capable of receiving various information about the vehicle V, such as location information and information about the presence or absence of passengers, from the terminal device 3, and transmitting dispatch instructions, etc. to each terminal device 3. Furthermore, this server 2 may be capable of performing part of the driving diagnosis processing described below.

[0028] The terminal device 3 can be configured as a device installed in the vehicle V. Specifically, it can be configured as an on-board device 3A mounted on the vehicle V (e.g., a taxi V1), or a mobile communication terminal 3B owned by the driver and installed in the vehicle V (e.g., a ride-sharing service providing vehicle V2) when driving. As the mobile communication terminal 3B, an information terminal having a communication function such as a smartphone, a tablet terminal, or a mobile personal computer can be adopted. In this embodiment, an example is shown in which the terminal device 3 functions as a driving diagnosis device 10.

[0029] FIG. 2 is a block diagram showing a hardware configuration of an example of a driving diagnosis device according to the first embodiment of the present disclosure. In FIG. 2, the driving diagnosis device 10 is illustrated as being realized by an on-board device 3A mounted on a taxi V1. As shown in FIG. 2, the driving diagnosis device 10 according to the present embodiment can be realized by a computer. In detail, the driving diagnosis device 10 may include a CPU (Central Processing Unit) 20 as an example of a processor, a ROM (Read Only Memory) 22 and a RAM (Random Access Memory) 24 as examples of memory, a storage 26, a communication interface 28, and an input / output interface 30. These configurations may be connected to each other so as to be able to communicate with each other via an internal bus 32.

[0030] The CPU 20 may be a central processing unit capable of executing various programs and controlling each part. Specifically, the CPU 20 may be capable of reading various programs stored in the ROM 22 or the storage 26 and executing the programs using the RAM 24 as a working area. The CPU 20 may be capable of controlling each component constituting the driving diagnosis device 10 and performing various arithmetic processing according to the programs.

[0031] The ROM 22 may be capable of storing various programs and various data, and the RAM 24 may be capable of temporarily storing programs or data as a working area.

[0032] The storage 26 may be configured with a recording medium such as a hard disk drive (HDD) or a solid state drive (SSD), and may store various programs including an operating system and various data necessary for operating the driving diagnosis device 10. In this embodiment, the ROM 22 or the storage 26 may store programs and various data for performing driving diagnosis processing.

[0033] The communication interface 28 may be an interface capable of wireless communication so that the driving diagnosis device 10 can communicate with the server 2 and other devices via a network NW. This communication interface 28 may use communication standards such as CAN (Controller Area Network), Ethernet (registered trademark), LTE (Long Term Evolution), FDDI (Fiber Distributed Data Interface), or Wi-Fi (registered trademark).

[0034] The input / output interface 30 may be an interface for transmitting and receiving data between various components mounted on the vehicle V or the terminal device 3, which are necessary for carrying out the driving diagnosis process. The components electrically connected to the input / output interface 30 may be appropriately selected according to the contents of the driving diagnosis. In this embodiment, as shown in FIG. 2, the input / output interface 30 is exemplified by an input / output interface to which several components provided on the vehicle V, specifically, a camera 40, a GPS (Global Positioning System) 42, an input switch 44, an acceleration sensor 46, a steering angle sensor 48, a vehicle speed sensor 50, a monitor 52, and a speaker 54, are electrically connected. Note that the input / output interface 30 and the various components described above may be directly connected to each other or may be connected to each other via an ECU (Electronic Control Unit) that controls each component.

[0035] The camera 40 may be configured with an imaging means capable of imaging the outside of the vehicle V and the inside of the vehicle cabin. The camera 40 may be a well-known image sensor such as a CCD (Charge Coupled Device) or CMOS (Complementary Metal Oxide Semiconductor) image sensor. The camera 40 may include one or more external cameras that are installed outside the vehicle V and capable of imaging the surroundings of the vehicle V, for example, the area in front of the vehicle, and one or more internal cameras that are installed inside the vehicle cabin and are directed toward the driver or the rear seat to image the situation inside the vehicle cabin. The external camera can be used to recognize the distance between the vehicle and the preceding vehicle traveling in front of the vehicle, as well as the lane and traffic lights. The internal camera can be used to detect the driver's line of sight and the presence or absence of a passenger.

[0036] The GPS 42 may be a device that is mounted on the vehicle V and is capable of measuring the current position of the vehicle V. The GPS 42 may include an antenna (not shown) that receives signals from GPS satellites.

[0037] The input switch 44 may be provided at a position where the driver can operate it, such as an instrument panel, a center console, or a steering wheel, and may be used to realize various operations related to the driving diagnosis device 10. The switch 44 may be configured as a push button switch, a touch panel, or the like. The input switch 44 may include switches related to the driving operation of the vehicle V, such as a switch for turning on the turn signals and hazard lights and a switch for operating the horn, or switches related to the business of the taxi V1, such as a switch for the driver to switch between occupied and unoccupied status of the vehicle V and a lever for opening and closing the rear door. In addition, when the driving diagnosis device 10 is applied to the mobile communication terminal 3B, the input switch 44 may also include a touch panel for operating an application (for example, an application related to a ride sharing service) on the mobile communication terminal 3B.

[0038] The acceleration sensor 46 may be a sensor capable of detecting acceleration in any direction of the vehicle V, for example, in three directions, i.e., forward / backward, left / right, and up / down. The steering angle sensor 48 may be a sensor capable of detecting the steering angle of the vehicle V, more specifically, the steering angle of the steering wheel of the vehicle V. The vehicle speed sensor 50 may be a sensor capable of detecting the vehicle speed of the vehicle V. Sensors having well-known structures can be adopted for these three sensors, and therefore a description of their specific configurations will be omitted.

[0039] The acceleration data detected by the acceleration sensor 46 can be used, for example, to determine whether or not the vehicle V has suddenly accelerated or decelerated (specifically, sudden acceleration or sudden braking). The steering angle data detected by the steering angle sensor 48 can be used, for example, to determine whether or not the vehicle V has suddenly accelerated or decelerated or made a U-turn. The vehicle speed data detected by the vehicle speed sensor 50 can be used, for example, to determine whether or not the vehicle V has suddenly accelerated or decelerated or whether or not the vehicle V has exceeded the legal speed limit or exceeded the speed limit.

[0040] The monitor 52 may be provided in a position visible to the driver at the front of the vehicle cabin, such as an instrument panel or a meter panel, and may be configured as a display capable of displaying various information. A liquid crystal panel, an organic EL panel, or the like may be used as the monitor 52. A display having a touch panel function may be used as a center display, which is an example of the monitor 52 provided on the instrument panel.

[0041] The speaker 54 may be provided at an appropriate location in the vehicle interior, such as on the dashboard or the front door, and may be capable of outputting audio to the driver. The speaker 54 and the above-mentioned monitor 52 may function as a notification unit 66 (see FIG. 3) that notifies the driver of the diagnosis result of the driving diagnosis device 10.

[0042] 3 is a block diagram showing a functional configuration of an example of a driving diagnosis device according to the first embodiment of the present disclosure. The driving diagnosis device 10 according to the present embodiment can realize various functions by using the above-mentioned hardware resources. In detail, as shown in FIG. 3, the driving diagnosis device 10 includes at least an acquisition unit 60 that acquires driving data indicating the driving state of the vehicle V, a driving evaluation unit 62 that evaluates the driving operation of the driver, a passenger presence / absence detection unit 64 that detects the presence or absence of a passenger in the vehicle V, and a notification unit 66 that notifies the driver of the driving evaluation result by the driving evaluation unit 62.

[0043] The acquisition unit 60 is for acquiring driving data indicating the driving state of the vehicle V. This acquisition unit 60 can be realized mainly by the above-mentioned input / output interface 30. The driving data acquired by the acquisition unit 60 may include various data of the vehicle V during driving. Specifically, this driving data may include, for example, imaging data captured by the camera 40, position information data acquired by the GPS 42, operation data of the input switch 44, and at least a portion of output signal data of various sensors (acceleration sensor 46, steering angle sensor 48, and vehicle speed sensor 50).

[0044] The driving evaluation unit 62 is for evaluating the driving operation of the driver based on the driving data acquired by the acquisition unit 60. This driving evaluation unit 62 can be mainly realized by the above-mentioned CPU 20. The evaluation of the driving operation may be, in particular, an evaluation of whether the driver is driving dangerously or driving in a manner that deteriorates fuel efficiency, or whether there is a risk of such driving. Specifically, the driving evaluation unit 62 may detect whether the driving data acquired by the acquisition unit 60 meets the conditions of a plurality of evaluation items (hereinafter also referred to as "evaluation item group") related to the driving operation set in advance, and evaluate the driving operation of the driver based on the number of times the detection is made.

[0045] FIG. 4 is a diagram showing an example of an evaluation item group (first evaluation item group) used in the driving evaluation unit shown in FIG. 3. All of the evaluation items (evaluation item group) used in the driving evaluation unit 62 according to this embodiment can be specified based on data obtainable via the input / output interface 30. In other words, the evaluation item group used in the driving evaluation unit 62 is appropriately set among items that can be evaluated by data obtainable via the input / output interface 30. As an example of the evaluation items constituting the evaluation item group, for example, as shown in FIG. 4, sudden acceleration, sudden braking, sudden steering, lane departure, distracted driving, insufficient distance between vehicles, and inadequate turn signal operation (specifically, delayed turn signal operation or turning right or left without turn signal operation) can be mentioned. Note that the above-mentioned evaluation items are merely examples, and even if other evaluation items are included, some of these evaluation items may not be adopted. Incidentally, evaluation items other than the above-mentioned evaluation items include, for example, exceeding the legal speed limit or speed limit, running beyond a stop line, and violation of a stop sign.

[0046] Each evaluation item constituting the above-mentioned evaluation item group may be evaluated as follows. That is, sudden acceleration and sudden braking can be evaluated based on signals from the acceleration sensor 46 or the vehicle speed sensor 50. Sudden steering can be evaluated based on a signal from the steering angle sensor 48. Lane departure and insufficient distance between vehicles can be evaluated based on an image of the front of the vehicle captured by the camera 40, particularly the external camera. Inattentive driving can be evaluated based on an image captured by the camera 40, particularly the internal camera facing the driver. And, defective turn signal operation can be evaluated based on the input switch 44, particularly the switch that operates the turn signal, and position information from the GPS 42 or a signal from the steering angle sensor 48.

[0047] In addition, the evaluation of the above-mentioned evaluation items can be performed using, for example, any threshold value. Specifically, for example, evaluation of sudden acceleration and sudden braking can be performed based on the number of times that the acceleration sensor 46 detects a signal indicating that acceleration or deceleration has occurred at or above a preset threshold value. In this embodiment, when it is detected that the number of times that the acceleration sensor 46 detects a signal indicating that acceleration or deceleration has occurred at or above the threshold value exceeds a specified number (hereinafter, also referred to as the "warning number"), a warning is issued to the driver via the notification unit 66. The warning number may be set for each evaluation item, and may be adjusted according to the contents of the evaluation item, for example, as shown in FIG. 4. In addition, the result of the evaluation by the driving evaluation unit 62 may be transmitted to the server 2 via the communication interface 28. The evaluation result transmitted to the server 2 can be used for driving management of the driver, etc.

[0048] The passenger presence / absence detection unit 64 is for detecting the presence or absence of a passenger in the vehicle V. Although a specific method for detecting the presence or absence of a passenger is not particularly limited, in this embodiment, a case where the presence or absence of a passenger is detected by using information that can directly detect the presence or absence of a passenger among information input via the input / output interface 30 is illustrated. More specifically, the presence or absence of a passenger can be detected based on, for example, an input operation of the input switch 44 (specifically, a switch for switching between an occupied vehicle and an empty vehicle for the taxi V1 business) by the driver. Note that, as a method for detecting the presence or absence of a passenger according to this embodiment, in addition to the above-mentioned one using the input switch 44, for example, a method for detecting the presence or absence of a passenger can be cited, which detects the presence or absence of a passenger based on the imaging result of an internal camera that can image the rear seat among the cameras 40, or based on the detection result of a pressure sensor (not shown) that is provided in advance on the seat of the rear seat.

[0049] The notification unit 66 is for notifying the driver of the driving evaluation result by the driving evaluation unit 62. This notification unit 66 can be realized mainly by at least one of the monitor 52 and the speaker 54 described above. Various methods can be adopted for the method of notifying the driving evaluation result by the notification unit 66. For example, the monitor 52 may display a comprehensive evaluation of the driving operation as a score or a message, or may display advice or a warning message related to the evaluation result, or the speaker 54 may output a message corresponding to the driving evaluation result, a warning sound, or the like.

[0050] Here, when the notification unit 66 notifies the driver of the driving evaluation result, audio output from the speaker 54 or message display on the monitor 52 can be used as described above, but making such a notification when there are passengers in the vehicle V may result in unnecessary information being conveyed to the passengers, which may hinder the passengers' comfortable travel. In particular, conveying to passengers a notification (hereinafter referred to as a "warning notification") that is implemented when the evaluation item evaluated by the driving evaluation unit 62 described above reaches a prescribed number of warnings may cause anxiety about the driver's driving operation and unnecessarily reduce the level of trust in the driver.

[0051] In the notification unit 66 according to this embodiment, taking the above points into consideration, when notifying the driver of the driving evaluation result, the notification method is changed based on the detection result of the passenger presence / absence detection unit 64. By changing the notification method, it is intended to perform driving diagnosis taking into consideration the situation inside the vehicle. More specifically, by changing the notification method so that the contents of the driving diagnosis result are not conveyed to passengers, it is intended to perform driving diagnosis without disturbing passengers' comfortable travel.

[0052] In this embodiment, the method of notifying the driving evaluation result by the notifying unit 66 when there is no passenger on board should be a method that allows the notification content to be clearly understood. Specifically, for example, the above-mentioned voice output by the speaker 54 and message display on the monitor 52 may be used alone or in combination. On the other hand, the method of notifying the driving evaluation result by the notifying unit 66 when there is no passenger on board should be a method that allows the notification content to be incomprehensible alone, or the notification unit 66 should not notify the driver when there is a passenger on board.

[0053] As a method of notifying the driving evaluation result by the notification unit 66, a method in which the notification content cannot be understood by itself may be, for example, only displaying an icon or a simple symbol mark (e.g., an exclamation mark) on the monitor 52. In addition, when the notification unit 66 does not notify the driver, the warning notification that should have been notified by the notification unit 66 may be at least temporarily stored in a memory area of ​​the storage 26 or the server 2, and the notification may be re-issued when there are no passengers on board, or may be made viewable at any time. Note that the method of notifying the driving evaluation result by the notification unit 66 taking into account the presence or absence of passengers is not limited to the above-mentioned one, and may be appropriately changed within a range in which the purpose can be achieved.

[0054] As described above, if the method of notifying the driving diagnosis result by the notifying unit 66 is changed depending on whether or not there is a passenger, a notification method that takes into account the presence or absence of a passenger can be adopted. In particular, if a method that does not transmit the notification content to the passenger is adopted as the notification method when there is a passenger, the passenger in the vehicle V will not substantially become aware of the notification of the driving diagnosis result, and the notification will not hinder the passenger's comfortable travel. In addition, since the driving diagnosis itself can be performed regardless of the presence or absence of a passenger, the driver can improve his driving technique and the quality of the passenger transportation service based on the driving diagnosis result.

[0055] Meanwhile, in commercial vehicles, driving is often performed according to passenger requests while the vehicle is in operation. For example, when passengers are in a hurry, the driver may drive at a higher vehicle speed or acceleration than usual, and when driving according to passenger directions, it is expected that sudden lane changes or right / left turns may occur due to delays in passenger directions. It is also known that commercial vehicle drivers tend to drive comfortably when the vehicle is in operation, and conversely, tend to drive more absent-mindedly when the vehicle is empty compared to when the vehicle is in operation. Therefore, driving operations in commercial vehicles can change significantly depending on whether the vehicle is in operation or empty.

[0056] Considering the above points, in the driving diagnosis device 10 according to this embodiment, in addition to changing the notification method by the notification unit 66 based on the presence or absence of a passenger, the evaluation method of the driving evaluation unit 62 may be changed based on the presence or absence of a passenger. Specifically, as described below, for example, the evaluation items evaluated by the driving evaluation unit 62 may be changed based on the presence or absence of a passenger, thereby realizing driving evaluation according to the occupied / empty vehicle state.

[0057] Fig. 5 is a diagram showing an example of the second evaluation item group of the driving evaluation unit shown in Fig. 3. The driving evaluation unit 62 according to this embodiment may perform driving evaluation taking into consideration the presence or absence of passengers by differentiating the evaluation item group evaluated when there are no passengers on board (hereinafter referred to as the "first evaluation item group") from the evaluation item group evaluated when there are passengers on board (hereinafter referred to as the "second evaluation item group"). In the following description, the evaluation item group described above with reference to Fig. 4 is an example of the first evaluation item group.

[0058] The first evaluation item group and the second evaluation item group may have different types and numbers of evaluation items, as shown in Figures 4 and 5. As a specific example, the first evaluation item group may be composed of seven items, namely, sudden acceleration, sudden braking, sudden steering, running out of the lane, distracted driving, insufficient distance between vehicles, and inadequate operation of a turn signal, whereas the second evaluation item group may be composed of four items, namely, sudden braking, sudden steering, running out of the lane, and inadequate operation of a turn signal. In the above example, the reason why the second evaluation item group does not include the three evaluation items, namely, sudden acceleration, insufficient distance between vehicles, and inadequate operation of a turn signal, is that it is possible to assume that the driver will intentionally perform these actions when the passenger is in a hurry or when the passenger instructs the driver to make a sudden right or left turn.

[0059] As described above, by switching between two evaluation item groups in the driving evaluation unit 62 and evaluating them, intentional sudden acceleration or improper turn signal operation due to passenger instructions will not be reflected in the driving diagnosis result, and a driving diagnosis result that is in line with the driver's own driving tendencies can be obtained. Note that the contents of the evaluation items constituting the first and second evaluation item groups described above are merely examples, and the evaluation items constituting each evaluation item group can be changed as appropriate. In addition, in this embodiment, an example is shown in which the evaluation items are changed based on the presence or absence of a passenger, but instead of or in addition to changing the evaluation items, the set number of warnings may be changed.

[0060] In the above-described embodiment, the case where the terminal device 3 constituting the driving diagnosis device 10 includes each component necessary for realizing the driving diagnosis process described below has been exemplified, but the driving diagnosis process may be realized by the driving diagnosis system 1 as a whole by including some of these components in the server 2. Specifically, at least a part of the driving evaluation unit 62, the passenger presence / absence detection unit 64, or the acquisition unit 60 may be provided in the server 2. For example, when the driving evaluation unit 62 is provided in the server 2, various driving data acquired by the acquisition unit 60 is sequentially transmitted to the server 2 via the communication interface 28, and the driving evaluation unit 62 in the server 2 performs driving evaluation of the vehicle V based on the driving data transmitted from the terminal device 3, and returns the result to the terminal device 3 at any timing.

[0061] Next, a brief description will be given of an example of a driving diagnosis process by the driving diagnosis device 10 including the above-mentioned series of configurations. The driving diagnosis process described below may be performed by the CPU 20 executing a program stored in the ROM 22 or the storage 26.

[0062] FIG. 6 is a flowchart showing an example of a driving diagnosis process by the driving diagnosis device according to the first embodiment of the present disclosure. The driving diagnosis process by the driving diagnosis device 10 according to this embodiment can be realized by generally carrying out the steps shown in FIG. 6. Specifically, first, when the operation of the vehicle V, specifically the taxi V1, is started by operating the ignition switch or the like, acquisition of driving data is started (step S11). The acquisition of driving data can be realized by acquiring signals from various sensors and the like mounted on the vehicle V via the input / output interface 30, as shown in FIG. 2.

[0063] After the above-mentioned step S11 or at the same timing as step S11, the passenger presence / absence detection unit 64 starts detecting the presence / absence of a passenger (step S12), and the driving evaluation unit 62 also starts driving evaluation (step S13). The passenger presence / absence detection unit 64 may be one that judges the presence / absence of a passenger based on the presence / absence of an input signal from the input switch 44. The driving evaluation unit 62 may be one that counts the number of times that a plurality of evaluation items set in advance exceed a threshold value set for each item, and evaluates driving based on whether or not the number of times exceeds a prescribed warning count. In addition, the driving evaluation unit 62 may switch between the above-mentioned first and second evaluation item groups and perform evaluation based on the detection result of the passenger presence / absence detection unit 64. The relative start timings of the above-mentioned steps S11 to S13 are not limited to the above-mentioned timings.

[0064] When the taxi V1 starts operating, the acquisition unit 60 continuously acquires driving data, and a driving evaluation is performed based on the acquired driving data. In parallel with this, the presence or absence of passengers is continuously detected. Then, when an event that should be notified to the driver as a result of the driving evaluation occurs (hereinafter, this will be referred to as a "notification event") (Yes in step S14), the notification unit 66 specifies a notification method for the notification event. The notification event may include a warning notification that is performed when the number of warnings set for each evaluation item shown in FIG. 4 or FIG. 5 described above is exceeded.

[0065] In step S14, when the occurrence of a notification event is detected, the passenger presence / absence detection unit 64 checks whether or not a passenger is present (step S15). If the check detects that no passenger is present, the notification unit 66 notifies the driver in a manner that allows the driver to clearly understand the contents of the notification event that has occurred (step S16). Specifically, the notification unit 66 may notify the driver by using both a message display on the monitor 52 and a voice output from the speaker 54. In this case, for example, a message such as "Avoid sudden acceleration" may be used as the message display. For example, the above-mentioned message may be output as a voice output. By performing such a notification, the driver can be aware of the diagnosis result of his / her driving in a timely manner.

[0066] If it is detected as a result of the check in step S15 that a passenger is on board, the notification unit 66 does not notify the driver of the generated notification event in order to avoid communicating the event to the passenger (step S17). Here, the notification event for which the driver has not been notified may be stored at least temporarily in the storage 26. The temporarily stored notification event may be made available for the driver to view at an appropriate time, such as when there is no passenger on board, or the driver may be notified again. When steps S16 and S17 are completed, the driving diagnosis device 10 returns to step S14 and continues to monitor for the occurrence of a new notification event.

[0067] As described above, the driving diagnosis device 10 or driving diagnosis system 1 according to this embodiment can change the method of notifying the driving diagnosis result depending on whether or not there are passengers, thereby making it possible to notify the driving diagnosis according to the situation inside the vehicle. Also, as described above, if the driving diagnosis result is not communicated to passengers when there are no passengers on board, it is possible to improve the driver's driving operation while suppressing the deterioration of the quality of passenger transportation services. Therefore, it is expected to have the effects of both improving passenger comfort and suppressing accidents.

[0068] <Second embodiment> When the driving diagnosis device 10 is realized in the vehicle-mounted device 3A mounted on the taxi V1 as in the first embodiment described above, it is relatively easy for the vehicle-mounted device 3A to acquire signals from various sensors provided in the vehicle V due to its structure, so that the presence or absence of a passenger can be detected easily and with high accuracy. On the other hand, depending on the embodiment of the driving diagnosis device, it may not be easy to acquire various data necessary for the driving diagnosis described above. For example, when the driving diagnosis device is realized in the form of an application in the mobile communication terminal 3B, in order to detect the presence or absence of a passenger, it may take a considerable amount of time and effort to install the device and to set up to enable data exchange between applications to acquire image data from the camera 40 (particularly the internal camera) provided in the vehicle V with the mobile communication terminal 3B, or to acquire information on an active vehicle / vacant vehicle operated by the driver in the application of the ride-sharing service. In addition, it is not realistic to require the driver to input information on an active vehicle / vacant vehicle just for driving diagnosis, because it would be an obstacle to transportation operations. Therefore, if a series of processes, including detecting the presence or absence of passengers, could be performed by the driving diagnosis device alone, without using various components installed in the vehicle V or operational information from other applications, and without requiring the driver to perform any input operations, this would simplify the installation and setup work of the device, and would be desirable.

[0069] Therefore, as a second embodiment of the present disclosure, a driving diagnosis device 10A capable of detecting the presence or absence of a passenger with high accuracy without taking into account data acquired specifically for the detection will be described below. In the following, a case where the driving diagnosis device 10A is realized by a mobile communication terminal 3B installed in a ride-sharing service providing vehicle V2 in the driving diagnosis system 1 shown in FIG. 1 will be described as an example. However, as already mentioned, the mobile communication terminal 3B is a terminal device to which the driving diagnosis device 10 according to the first embodiment described above can also be applied, and is not limited to the aspect described below. Similarly, the driving diagnosis device 10A according to this embodiment can also be realized by an on-board device 3A installed in a taxi V1.

[0070] The driving diagnosis device 10A according to this embodiment may have a configuration similar to that of the driving diagnosis device 10 according to the first embodiment described above. Therefore, the following mainly describes the configuration different from that of the driving diagnosis device 10 according to the first embodiment described above, and the same structures as those of the driving diagnosis device 10 according to the first embodiment described above are given the same reference numerals and their description is omitted. Furthermore, the driving diagnosis device 10A described below may be configured to realize driving diagnosis processing as a driving diagnosis system by including some of its components in the server 2, similar to the driving diagnosis device 10 according to the first embodiment.

[0071] Fig. 7 is a block diagram showing a hardware configuration of an example of a driving diagnosis device according to the second embodiment of the present disclosure. The driving diagnosis device 10A according to the present embodiment can be applied to, for example, a mobile communication terminal 3B provided in a ride-sharing service providing vehicle V2 shown in Fig. 1. In addition, as shown in Fig. 7, the hardware configuration of this driving diagnosis device 10A may include a CPU 20 as an example of a processor, a ROM 22 and a RAM 24 as examples of memories, a storage 26, a communication interface 28, and an input / output interface 30A, similar to the driving diagnosis device 10 according to the first embodiment described above.

[0072] The input / output interface 30A of the present embodiment may be an interface for transmitting and receiving data between various components in the mobile communication terminal 3B, which are necessary for carrying out the driving diagnosis process. That is, the input / output interface 30A of the present embodiment may not be electrically connected to the vehicle V. As components electrically connected to the input / output interface 30A, for example, as shown in FIG. 7, a GPS 42, an input switch 44, an acceleration sensor 46, a monitor 52, and a speaker 54 may be electrically connected.

[0073] Fig. 8 is a block diagram showing a functional configuration of an example of a driving diagnosis device according to a second embodiment of the present disclosure. The driving diagnosis device 10A according to this embodiment can realize various functions by using the above-mentioned hardware resources. In detail, as shown in Fig. 8, the driving diagnosis device 10A includes an acquisition unit 60 that acquires driving data indicating the driving state of the vehicle V, a driving evaluation unit 62 that evaluates the driving operation of the driver, a passenger presence / absence detection unit 64A that detects the presence or absence of a passenger in the vehicle V, and a notification unit 66 that notifies the driver of the driving evaluation result by the driving evaluation unit.

[0074] The acquisition unit 60, driving evaluation unit 62, and notification unit 66 described above may have substantially the same functions as those of the driving diagnosis device 10 according to the first embodiment. On the other hand, the passenger presence / absence detection unit 64A of the driving diagnosis device 10A according to the present embodiment may further include an inference unit 68 that infers the presence or absence of a passenger in the vehicle V from the driving data acquired by the acquisition unit 60. That is, the passenger presence / absence detection unit 64A of the driving diagnosis device 10A according to the present embodiment attempts to identify the presence or absence of a passenger by inferring it using the inference unit 68, rather than referring to a signal or the like that directly indicates the presence of a passenger.

[0075] The inference unit 68 may estimate the presence or absence of a passenger by using a trained model obtained by machine learning. The trained model referred to here may be configured as a model that has been trained by machine learning so that the presence or absence of a passenger can be estimated from the travel data, and may be an inference model trained using various machine learning methods such as a neural network model, a decision tree model, or a support vector machine. When the travel data acquired by the acquisition unit 60 is input to the input layer of this trained model, an estimation result of the presence or absence of a passenger may be obtained.

[0076] The trained model used in the inference unit 68 can be obtained by using the machine learning device 70. Hereinafter, the machine learning device 70 for obtaining the trained model and a method for generating the trained model will be described.

[0077] FIG. 9 is a block diagram showing a functional configuration of an example of a machine learning device according to the second embodiment of the present disclosure. The machine learning device 70 according to this embodiment may be realized by the server 2. As shown in FIG. 9, this machine learning device 70 may include a learning dataset acquisition unit 74, a learning dataset storage unit 76, a learning processing unit 78, and a trained model storage unit 80. Note that, although the machine learning device 70 is realized by the server 2 in this embodiment, this machine learning device 70 can also be realized by the terminal device 3. In that case, the terminal device 3 alone can generate a trained model and use it to estimate the presence or absence of a passenger.

[0078] The learning dataset acquisition unit 74 is connectable to the driving diagnosis device 10A via the network NW, for example, and acquires a plurality of data constituting a learning (training) dataset transmitted from the terminal device 3 via the communication interface of the server 2. The learning dataset can be a dataset composed of input data (sometimes called "explanatory variables") including driving data indicating the driving state of the vehicle V, and output data (sometimes called "objective variables") including data indicating the presence or absence of a passenger in the vehicle V. Therefore, the data acquired by the learning dataset acquisition unit 74 may include at least two types of data: driving data for an arbitrary period acquired by the acquisition unit 60 of the driving diagnosis device 10A, and data indicating the presence or absence of a passenger at that time. The driving data can be data acquired by the acquisition unit 60 of the driving diagnosis device 10A, specifically, an output signal of the acceleration sensor 46 or position information measured by the GPS 42 for an arbitrary period, but is not limited to these as long as it is data related to the driving of the vehicle V that can be acquired by the acquisition unit 60. Furthermore, data indicating the presence or absence of a passenger constituting one learning dataset may be a single value (that is, a value corresponding to either "presence" or "absence").

[0079] The travel data and the data indicating the presence or absence of passengers acquired by the learning dataset acquisition unit 74 may be associated with each other as one learning dataset beforehand or after acquisition by the learning dataset acquisition unit 74. A plurality of sets of these data may be generated for each vehicle V or each driver.

[0080] The learning dataset storage unit 76 may be a recording medium such as a database for storing various data acquired by the learning dataset acquisition unit 74 in units of learning datasets for each vehicle V or for each driver. Note that the learning dataset storage unit 76 may be omitted. In that case, the learning datasets acquired by the learning dataset acquisition unit 74 may be sequentially sent to the learning processing unit 78.

[0081] The learning processing unit 78 is for generating a learned model by learning a learning model using a plurality of sets of learning datasets stored in the learning dataset storage unit 76. In this embodiment, as described below, supervised learning using a neural network is adopted as a specific method of machine learning. As described above, the specific method of machine learning is not limited to this, and other learning methods can be adopted as long as they can learn the correlation between input and output from the learning dataset.

[0082] As shown in FIG. 9, the neural network model trained in the learning processing unit 78 may include one or more neurons constituting an input layer to which input data is input, one or more neurons constituting an output layer, a plurality of neurons constituting an intermediate layer provided between the input layer and the output layer, and nodes connecting the neurons. The number of neurons constituting the input layer and the output layer may be adjusted according to the number of pieces of input data and output data. The number of layers of the intermediate layer (sometimes called a "hidden layer") may be one as illustrated in FIG. 9, but is preferably set to two or more. An arbitrary weight value may be associated with the nodes stretched between the neurons constituting each layer. The learning processing unit 78 can generate a trained model by training the neural network model having the above-mentioned configuration using a training data set. A specific method for generating a trained model will be described later.

[0083] The trained model storage unit 80 may be a database for storing the trained model generated by the learning processing unit 78. The trained model stored in the trained model storage unit 80 is applied to a real system, that is, an arbitrary driving diagnosis device 10A, in response to a request from an administrator of the vehicle V. In FIG. 9, the training data set storage unit 76 and the trained model storage unit 80 are illustrated as separate storage means for convenience of explanation, but they may be configured by a single recording medium (database). The trained model storage unit 80 may also be omitted, similar to the training data set storage unit 76. In that case, the trained models trained and generated by the learning processing unit 78 may be sequentially transmitted to the corresponding driving diagnosis device 10A.

[0084] Next, a method for generating a trained model used in the passenger presence detection unit 64A according to this embodiment will be described. In the following description, a method for generating a trained model by performing supervised learning using a training dataset (sometimes called teacher data) using the above-mentioned machine learning device 70 will be exemplified. In this regard, the following description of effects and the like also serves as a description of the effects of the machine learning device 70 according to this embodiment. Note that the trained model generation method of the present disclosure may be performed by any computer, and is not limited to the machine learning device 70. In addition, the trained model generation method according to this embodiment may be provided in the form of a program that causes a processor of a computer to execute a predetermined operation, or in the form of a non-volatile computer-readable medium storing the program.

[0085] FIG. 10 is a flowchart showing an example of a trained model generation method according to the second embodiment of the present disclosure. When generating a trained model, first, a plurality of training datasets to be used for training are acquired in the training dataset acquisition unit 74 (step S21). As described above, the training dataset acquired here may be composed of input data composed of driving data indicating the driving state of an arbitrary vehicle V (for example, the ride-sharing service providing vehicle V2 shown in FIG. 1) and output data composed of data indicating the presence or absence of a passenger in the vehicle V. In addition, the number of training datasets acquired in step S21 may be determined based on the number assumed to be necessary for generating a trained model capable of obtaining an inference result with a desired accuracy, for example. The training dataset acquired here may be stored in the training dataset storage unit 76.

[0086] Next, a learning model for learning is prepared (step S22). The learning model prepared here can be a model before learning is performed, for example, a neural network model in which the weight values ​​set for each node are set to arbitrary initial values.

[0087] Once the acquisition of the learning datasets used for learning and the preparation of the learning model for learning are completed, learning is then performed by the learning processing unit 78. The processing performed here may be processing using a plurality of sets of learning datasets acquired in step S21 to train a learning model that estimates a correlation between driving data (explanatory variables) indicating the driving state of the vehicle V and the presence or absence of a passenger inside the vehicle V (objective variable). As a specific method of learning, for example, a method of executing the processing shown in the following steps S23 to S26 can be adopted.

[0088] When learning of the learning model is started, first, one learning dataset to be used for learning is selected from the multiple sets of learning datasets acquired by the learning dataset acquisition unit 74 (step S23). Next, the input data in this selected one learning dataset, i.e., the running data, is input to the neurons in the input layer of the learning model prepared in step S21 (step S24). When the running data is input to the neurons in the input layer, an arbitrary value is output to the neurons in the output layer of the learning model. This output may be calculated using a general method of calculating the output value of a neural network, that is, a method of calculating the value of the output neuron as the sum of a sequence of multiplication values ​​of the value of the input neuron connected to the neuron and the weight value associated with the node connecting the output neuron and the input neuron, for all neurons other than the neurons in the input layer.

[0089] In the above-mentioned step S24, when inputting the input data of the learning dataset into the neurons of the input layer, the format in which the input data is input into the neurons of the input layer may be appropriately set in consideration of the accuracy of the trained model to be generated, etc. For example, the driving data as the input data may be input as explanatory variables as it is, as the data acquired by the learning dataset acquisition unit 74 (specifically, the output signal of the acceleration sensor 46 and the position information measured by the GPS 42, etc.), or information on evaluation items such as sudden acceleration, sudden braking, and sudden steering may be specified from the above-mentioned output signal of the acceleration sensor 46 and the position information measured by the GPS 42 (such processing may be referred to as "preprocessing"), and this information may be input into the neurons of the input layer as explanatory variables.

[0090] When an arbitrary value is output to the neuron in the output layer of the learning model in step S24, the learning model is then trained using the output data in the one learning dataset selected in step S23, i.e., data on the presence or absence of passengers (step S25). For the training here, for example, a method can be adopted in which the value output to the neuron in the output layer of the learning model in step S24 is compared with information on the presence or absence of passengers constituting the output data of the one learning dataset described above to find an error, and the weight value associated with each node is adjusted using backpropagation so as to reduce the found error.

[0091] When the weight values ​​of the learning model are adjusted in step S25, it is determined whether or not the machine learning can be terminated. If learning is to be continued (No in step S26), the process returns to step S23, a new learning dataset is selected, and the above-described series of steps is repeated.

[0092] When a predetermined condition is met, such as when a series of steps shown in steps S23 to S26 are repeated a predetermined number of times or when the error becomes smaller than a preset allowable value, it is determined in step S26 that the machine learning can be ended (Yes in step S26). In this case, the learning model obtained after the series of learning is stored in the learned model storage unit 80 as a trained model (step S27), and the series of the learning model generation method is completed.

[0093] As described above, according to the machine learning device 70 and the method for generating a trained model according to this embodiment, it is possible to obtain a trained model that has learned the correlation between the traveling data of the vehicle V and the presence or absence of a passenger in the vehicle V. The trained model obtained here is used by the inference unit 68 in the passenger presence / absence detection unit 64A in the driving diagnosis device 10A of the vehicle V, thereby enabling the driving diagnosis device 10A alone to detect the presence or absence of a passenger.

[0094] Next, an example of driving diagnosis processing by the driving diagnosis device 10A including the inference unit 68 that uses the above-mentioned trained model will be briefly described. The driving diagnosis processing described below may be performed by the CPU 20 executing a program stored in the ROM 22 or the storage 26, similar to that in the driving diagnosis device 10 according to the first embodiment.

[0095] Fig. 11 is a flowchart showing an example of a driving diagnosis process by the driving diagnosis device according to the second embodiment of the present disclosure. As shown in Fig. 11, the driving diagnosis process by the driving diagnosis device 10A according to this embodiment can be the same as the process shown in Fig. 6, except that the presence or absence of a passenger is identified based on an estimation result by the inference unit 68 in the passenger presence / absence detection unit 64A. Therefore, the details of the processing steps described below are appropriately cited from those described in the first embodiment, and the description thereof will be omitted.

[0096] Specifically, first, when the operation of the ride-sharing service by the ride-sharing service providing vehicle V2 is started by starting the driving diagnosis device 10, etc., acquisition of driving data is started (step S11), and thereafter, or at the same timing, the passenger presence / absence detection unit 64A starts estimating the presence or absence of a passenger (step S12A), and the driving evaluation unit 62 starts driving evaluation (step S13). The passenger presence / absence detection unit 64A can estimate the presence or absence of a passenger by inputting the driving data acquired by the acquisition unit 60 to the input layer of the trained model in the inference unit 68.

[0097] Generally, passengers get on and off the vehicle V when the vehicle V is stopped for a certain period of time. Therefore, it is considered that the detection result of the presence or absence of a passenger in the vehicle V will not change unless the vehicle V is stopped for a certain period of time. Therefore, it is preferable that the passenger presence / absence detection unit 64A estimates the presence or absence of a passenger every time a stoppage for a certain period of time is detected. By setting the estimation timing of the presence or absence of a passenger in this manner, it is possible to secure a relatively long period of traveling data as the traveling data input to the trained model at the time of estimation, and the estimation accuracy can be improved.

[0098] When the ride-sharing service providing vehicle V2 starts operation, the acquisition unit 60 continuously acquires driving data, and a driving evaluation is performed based on the acquired driving data. Then, when a notifiable event occurs (Yes in step S14), the notification unit 66 specifies a method of notifying the notifiable event. Specifically, it is confirmed whether or not the passenger presence / absence detection unit 64A estimates that there is a passenger (step S15). As a result, when it is estimated that there is no passenger, the notification unit 66 notifies the driver in a manner that allows the driver to clearly understand the contents of the notifiable event that has occurred (step S16). On the other hand, when it is estimated that there is a passenger, the notification unit 66 does not notify the driver to avoid the passenger from being informed of the notifiable event that has occurred (step S17). When steps S16 and S17 are completed, the driving diagnosis device 10 returns to step S14 and continues to monitor the occurrence of a new notifiable event.

[0099] As described above, in the driving diagnosis device 10A according to the present embodiment, the method of notifying the driving diagnosis result can be changed depending on the presence or absence of a passenger, thereby making it possible to notify the driving diagnosis according to the situation inside the vehicle. Furthermore, as described above, if the driving diagnosis result is not communicated to passengers when there are no passengers on board, it is possible to improve the driver's driving operation while suppressing the deterioration of the quality of passenger transportation services. Furthermore, since the presence or absence of a passenger can be estimated by using the inference unit 68, a series of driving diagnosis processes can be realized by the driving diagnosis device alone without using signals from components mounted on the vehicle V that can directly detect the presence or absence of a passenger or operation information of other applications, and without requesting the driver to perform an input operation. This simplifies the installation and setting work of the driving diagnosis device 10A.

[0100] In addition, the trained model described in the above embodiment is generated by performing learning by batch learning in the machine learning device 70, but the timing of learning is not limited to this. For example, the trained model generated through the above-mentioned batch learning may be further trained online to further improve the estimation accuracy. In that case, the driving data acquired by the acquisition unit 60 of the driving diagnosis device in which the trained model is implemented and the data indicating the presence or absence of a passenger at the time the driving data was acquired are temporarily stored in the storage 26 or the like as a set of online learning data sets, and the machine learning process of the trained model used in the inference unit 68 using the online learning data set may be executed at any timing. The data indicating the presence or absence of a passenger constituting the online learning data set may be data input by the driver to the terminal device 3 or data identified based on the operation data collected by the server 2.

[0101] The driving diagnosis system 1, the driving diagnosis device 10, 10A, the machine learning device 70, and the method for generating a trained model according to the embodiments of the present disclosure have been described above, but it goes without saying that various modifications are possible without departing from the scope of the present disclosure. For example, in the above embodiment, the sudden acceleration and sudden braking are detected based on the signal from the acceleration sensor 46 or the vehicle speed sensor 50, but this is not limiting. For example, electrical signals input from the accelerator pedal and the brake pedal may be obtained, and sudden acceleration and sudden braking may be detected based on the signals.

[0102] In the above embodiment, a notification is given when the event set as an evaluation item reaches a predetermined number of warnings, but the present invention is not limited to this. For example, a number of advance notifications less than the number of warnings may be separately set, and when the advance notification number is reached, the driver may be notified in a different manner (for example, a warning display on the monitor 52) from the case when the warning number is reached.

[0103] Furthermore, in the above embodiment, the notification process executed by the CPU 20 by reading the program stored in the ROM 22 or the storage 26 may be executed by various processors other than the CPU 20. In this case, examples of the processor include a PLD (Programmable Logic Device) such as an FPGA (Field-Programmable Gate Array) whose circuit configuration can be changed after manufacture, and a dedicated electric circuit such as an ASIC (Application Specific Integrated Circuit) which is a processor having a circuit configuration designed exclusively for executing a specific process. The driving evaluation process may be executed by one of these various processors, or may be executed by a combination of two or more processors of the same or different types, for example, by a plurality of FPGAs, or a combination of a CPU and an FPGA. The hardware configuration of these various processors may be an electric circuit combining circuit elements such as semiconductor elements.

[0104] Furthermore, in the above embodiment, the program for driving diagnosis executed by the CPU 20 is prestored in the ROM 22 or storage 26 in the terminal device 3, but the present invention is not limited thereto. For example, various data including the program used for driving diagnosis may be stored in a non-transient computer-readable medium, such as a recording medium separate from the terminal device 3, specifically a magnetic disk (e.g., HDD), an optical disk (e.g., CD (Compact Disk) or DVD (Digital Versatile Disk) or a flash memory (e.g., USB (Universal Serial Bus) memory), and the program may be executable by electrically connecting the data to the terminal device 3. Similarly, the program used for driving diagnosis may be provided in the form of an application from a server or the like via a network NW. [Explanation of symbols]

[0105] 1. Driving diagnostic system 2 Server 3 Terminal Equipment 10, 10A Driving diagnostic device 60 Acquisition Department 62 Driving Evaluation Department 64, 64A Passenger presence detection unit 66 Notification Department 68 Reasoning part 70 Machine Learning Device 74 Learning Data Set Acquisition Unit 78 Learning Processing Unit V Vehicle

Claims

1. An acquisition unit that acquires driving data indicating a driving state of a vehicle; a driving evaluation unit that evaluates a driving operation of a driver of the vehicle based on the traveling data acquired by the acquisition unit; a passenger presence detection unit for detecting the presence or absence of a passenger in the vehicle; a notification unit that notifies a driver of the driving evaluation result detected by the driving evaluation unit and changes a notification method of the driving evaluation result based on a detection result of the passenger presence / absence detection unit; The driving evaluation unit evaluates the driving operation of the driver based on the number of times that the driving data acquired by the acquisition unit meets the conditions of a plurality of evaluation items related to a preset driving operation, and evaluates the driving operation of the driver using a first evaluation item group used when the passenger presence / absence detection unit detects that there is no passenger, and a second evaluation item group different from the first evaluation item group used when the passenger presence / absence detection unit detects that there is a passenger. Driving diagnostic device.

2. An acquisition unit that acquires driving data indicating a driving state of a vehicle; a driving evaluation unit that evaluates a driving operation of a driver of the vehicle based on the traveling data acquired by the acquisition unit; a passenger presence detection unit for detecting the presence or absence of a passenger in the vehicle; a notification unit that notifies a driver of the driving evaluation result detected by the driving evaluation unit and changes a notification method of the driving evaluation result based on a detection result of the passenger presence / absence detection unit; The passenger presence detection unit includes an inference unit that infers the presence or absence of a passenger from the running data, The inference unit outputs an estimation result of the presence or absence of a passenger by inputting the traveling data to an input layer of a trained model that has undergone machine learning to estimate the presence or absence of a passenger. Driving diagnostic device.

3. A driving diagnosis system including a server and a terminal device installed in a vehicle, the server and terminal device being communicatively connected to each other via a network, An acquisition unit that acquires driving data indicating a driving state of a vehicle; a driving evaluation unit that evaluates a driving operation of a driver of the vehicle based on the traveling data acquired by the acquisition unit; a passenger presence detection unit for detecting the presence or absence of a passenger in the vehicle; a notification unit that is provided in the terminal device and notifies a driver of a driving evaluation result by the driving evaluation unit, and changes a notification method of the driving evaluation result based on a detection result of the passenger presence / absence detection unit; The driving evaluation unit evaluates the driving operation of the driver based on the number of times that the driving data acquired by the acquisition unit meets the conditions of a plurality of evaluation items related to a preset driving operation, and evaluates the driving operation of the driver using a first evaluation item group used when the passenger presence / absence detection unit detects that there is no passenger, and a second evaluation item group different from the first evaluation item group used when the passenger presence / absence detection unit detects that there is a passenger. Driving diagnostic system.

4. A driving diagnosis system including a server and a terminal device installed in a vehicle, the server and terminal device being communicatively connected to each other via a network, An acquisition unit that acquires driving data indicating a driving state of a vehicle; a driving evaluation unit that evaluates a driving operation of a driver of the vehicle based on the traveling data acquired by the acquisition unit; a passenger presence detection unit for detecting the presence or absence of a passenger in the vehicle; a notification unit that is provided in the terminal device and notifies a driver of a driving evaluation result by the driving evaluation unit, and changes a notification method of the driving evaluation result based on a detection result of the passenger presence / absence detection unit; The passenger presence detection unit includes an inference unit that infers the presence or absence of a passenger from the running data, The inference unit outputs an estimation result of the presence or absence of a passenger by inputting the traveling data to an input layer of a trained model that has undergone machine learning to estimate the presence or absence of a passenger. Driving diagnostic system.

5. A machine learning device for acquiring a trained model used in the inference unit of the driving diagnosis device according to claim 2, a learning dataset acquisition unit that acquires a plurality of learning datasets each including input data including running data indicating a running state of a vehicle and output data including data indicating the presence or absence of a passenger in the vehicle; A learning processing unit that uses a plurality of sets of the learning datasets acquired by the learning dataset acquisition unit to learn a learning model that estimates a correlation between the input data and the output data; A machine learning device comprising:

6. A method for generating a trained model used in the inference unit of the driving diagnosis device according to claim 2, comprising: Obtaining a plurality of learning data sets each including input data including driving data indicating a driving state of a vehicle and output data including data indicating the presence or absence of a passenger in the vehicle; A step of learning a learning model that estimates a correlation between the input data and the output data by using a plurality of the learning data sets; A method for generating a trained model comprising:

Images