Information processing system, information processing method, and program

Abstract

To provide an information processor, an information processing method, and a program that can provide more useful information to drivers.SOLUTION: An information processor includes a prediction unit that predicts the accident probability of a vehicle driven by a user, and an output control unit that outputs information to the user according to factors that have increased the accident probability as predicted by the prediction unit.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] This disclosure relates to information processing system The present invention relates to an information processing method and a program. [Background technology]

[0002] In recent years, attention has been focused on transportation systems that improve vehicle safety by using on-board sensors or roadside sensors such as surveillance cameras. In such transportation systems, useful information is extracted by integrating information detected by multiple sensors, and is used to notify the driver or as input information for automatic driving.

[0003] As an example of such technology, for example, Patent Document 1 listed below discloses a system that records points where a driver is presumed to have felt danger while driving a vehicle and issues a warning when the driver approaches those points. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2007-47914 A Summary of the Invention [Problem to be solved by the invention]

[0005] However, the technology disclosed in the above patent document merely provides a warning based on past records. Therefore, when the situation has changed since the time of the past record, the information provided to the driver may not be useful. Therefore, the present disclosure proposes a new and improved information processing device, information processing method, and program capable of providing more useful information to the driver. [Means for solving the problem]

[0006] According to the present disclosure, a generation unit generates a feature vector based on sensor information related to a vehicle or a driver, an acquisition unit acquires information related to locations where accidents are likely to occur by inputting the feature vector into a prediction model, and a prediction unit that acquires information related to the locations where accidents are likely to occur. and Warning for said location On the map An information processing system is provided that includes an output control unit that controls the output.

[0007] According to the present disclosure, an information processing system generates a feature vector based on sensor information related to a vehicle or a driver, inputs the feature vector into a prediction model to obtain information related to locations where accidents are likely to occur, and and Warning for said location On the map Control the output.

[0008] According to the present disclosure, a computer generates a feature vector based on sensor information related to a vehicle or a driver, inputs the feature vector into a prediction model to obtain information related to locations where accidents are likely to occur, and calculates the location. and Warning for said location On the map A program is provided that controls the output. Effect of the Invention

[0009] As described above, according to the present disclosure, it is possible to provide more useful information to the driver. Note that the above effects are not necessarily limited, and any of the effects shown in this specification or other effects that can be understood from this specification may be achieved in addition to or instead of the above effects. [Brief description of the drawings]

[0010] [Figure 1] FIG. 1 is an explanatory diagram for explaining an overview of a system according to an embodiment of the present invention. [Diagram 2] FIG. 2 is a block diagram showing an example of a logical configuration of the sensor device according to the present embodiment. [Diagram 3] FIG. 2 is a block diagram showing an example of a logical configuration of a server according to the present embodiment. [Figure 4] FIG. 11 is a diagram for explaining an example of a UI according to the embodiment. [Diagram 5] FIG. 11 is a diagram for explaining an example of a UI according to the embodiment. [Figure 6] FIG. 11 is a diagram for explaining an example of a UI according to the embodiment. [Figure 7] FIG. 11 is a diagram for explaining an example of a UI according to the embodiment. [Figure 8] FIG. 11 is a diagram for explaining an example of a UI according to the embodiment. [Figure 9] FIG. 11 is a diagram for explaining an example of a UI according to the embodiment. [Figure 10] FIG. 11 is a diagram for explaining an example of a UI according to the embodiment. [Figure 11] FIG. 11 is a diagram for explaining an example of a UI according to the embodiment. [Figure 12] FIG. 11 is a diagram for explaining an example of a UI according to the embodiment. [Figure 13] FIG. 11 is a diagram for explaining an example of a UI according to the embodiment. [Figure 14] FIG. 11 is a sequence diagram showing an example of the flow of a prediction model learning process executed in the system according to the present embodiment. [Figure 15] 10 is a flowchart showing an example of the flow of an accident probability prediction process executed in the sensor device according to the present embodiment. [Figure 16] 10 is a flowchart showing an example of the flow of a feature quantity vector generation process executed in the sensor device according to the present embodiment. [Figure 17] 10 is a flowchart showing an example of the flow of a feature quantity vector generation process related to accident information linked to a map, which is executed in the sensor device according to the present embodiment. [Figure 18] 10 is a flowchart showing an example of the flow of a feature quantity vector generation process relating to the occurrence probability of a predictive phenomenon, which is executed in the sensor device according to the present embodiment. [Figure 19] FIG. 1 is a block diagram showing an example of a schematic configuration of a vehicle system. [Figure 20]FIG. 2 is a block diagram showing an example of a hardware configuration of an information processing device. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] Hereinafter, preferred embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. In this specification and the drawings, components having substantially the same functional configurations are denoted by the same reference numerals, and redundant description will be omitted.

[0012] The explanation will be given in the following order. 1. Introduction 1.1. Overview 1.2.Technical issues 2. Configuration Example 2.1. Example of the configuration of the sensor device 100 2.2. Example of Server 30 Configuration 3. Technical features 4. Processing flow 5. Hardware configuration example 6. Summary

[0013] <<1. Introduction>> <1.1. Overview> First, an overview of a system 1 according to an embodiment of the present disclosure will be described with reference to FIG.

[0014] FIG. 1 is an explanatory diagram for explaining an overview of a system 1 according to this embodiment. FIG. 1 shows a situation in which a vehicle 10A is traveling in an X direction and a vehicle 10B is traveling in a -Y direction and is entering an intersection. For example, if the driver of the vehicle 10A cannot see the vehicle 10B, there is a possibility that the vehicles 10A and 10B will collide with each other. The same applies to the driver of the vehicle 10B. In a situation in which such an accident is predicted, it is desirable to provide appropriate information to the drivers of the vehicles 10A and 10B.

[0015] Therefore, the system 1 accumulates information detected by various sensor devices, predicts the probability of an accident, and provides the user with information based on the prediction results.

[0016] The sensor device (the sensor device 100 described with reference to FIG. 2) may be mounted on a moving object such as the vehicles 10A and 10B. The sensor device may also be mounted on a facility such as the traffic lights 20A and 20B. For example, the sensor device mounted on the vehicle 10A and the sensor device mounted on the traffic light 20A perform sensing on the vehicle 10A or the driver of the vehicle 10A. Similarly, the sensor device mounted on the vehicle 10B and the sensor device mounted on the traffic light 20B perform sensing on the vehicle 10B or the driver of the vehicle 10B. The system 1 can perform processing based on the accumulated sensor information by accumulating sensor information obtained as a result of such sensing. For example, the sensor device mounted on the vehicle 10A can provide information to the driver to avoid an accident with the vehicle 10B. The sensor information obtained by each sensor device may be shared within the system 1.

[0017] Accumulation of the sensor information and processing based on the accumulated sensor information may be performed in the server 30. The server 30 may be provided, for example, on a cloud. For reducing the latency of the service, the server 30 may be realized as a MEC (Mobile edge Computing) server provided, for example, in a mobile communication base station.

[0018] <1.2.Technical issues> Regarding the above-described system 1, the above-mentioned Patent Document 1 discloses a system that records a location where a driver is estimated to have felt danger while driving a vehicle based on information obtained by a sensor, and issues a warning when the driver approaches the location. However, in the system disclosed in Patent Document 1, when the situation has changed since the time when the information was recorded in the past, the information provided to the driver may not be useful. Furthermore, the driver may simply receive a warning, and may not be clear about what specific accident avoidance action should be taken.

[0019] Taking the above circumstances into consideration, the present invention has been developed to provide a system 1 according to an embodiment of the present disclosure. The system 1 according to the present embodiment is capable of predicting the accident probability and outputting information to a user according to factors that increase the accident probability.

[0020] As a mechanism for avoiding accidents, for example, a technology that measures the distance to other vehicles or pedestrians in the vicinity using a camera or a sensor, predicts a collision, and automatically stops the vehicle, etc. However, it is considered that there is a lot of information to be considered other than the distance. Therefore, the system 1 according to the present embodiment predicts the probability of an accident by utilizing various information that may affect the occurrence of an accident.

[0021] In addition, although the analysis of the cause of an accident is performed based on accident cases, the analysis may be difficult because the absolute number of accidents is small. Therefore, in the system 1 according to the present embodiment, the accident rate is predicted by taking into account accident precursor events.

[0022] <<2. Configuration Example>> 2.1. Example of the configuration of the sensor device 100 2 is a block diagram showing an example of a logical configuration of the sensor device 100 according to this embodiment. As shown in FIG. 2, the sensor device 100 includes a detection unit 110, a communication unit 120, an output unit 130, a storage unit 140, and a control unit 150.

[0023] (1) Detection unit 110 The detection unit 110 detects various information related to the vehicle, the interior of the vehicle, passengers (driver or passengers), the environment around the vehicle, etc. The detection unit 110 has various sensors such as a camera, a radar, an inertial sensor, or a biometric sensor, and outputs detected sensor information.

[0024] (2) Communications Department 120 The communication unit 120 is a communication module for transmitting and receiving data to and from other devices via wired / wireless communication. For example, the communication unit 120 communicates with other sensor devices 100 or the server 30.

[0025] (3) Output unit 130 The output unit 130 outputs information by video, image, sound, vibration, etc. For example, when the sensor device 100 is mounted on a vehicle, the output unit 130 outputs information to a passenger. When the sensor device 100 is mounted on a traffic light, the output unit 130 outputs information to a passenger of a surrounding vehicle or a pedestrian.

[0026] (4) Storage section 140 The memory unit 140 temporarily or permanently stores programs and various data for the operation of the sensor device 100 .

[0027] (5) Control unit 150 The control unit 150 provides various functions of the sensor device 100. The control unit 150 includes a prediction unit 152 and an output control unit 154. The control unit 150 may further include other components in addition to these components. That is, the control unit 150 may also perform operations other than those of these components. The operations of the prediction unit 152 and the output control unit 154 will be described in detail later.

[0028] <2.2. Example of Server 30 Configuration> 3 is a block diagram showing an example of a logical configuration of the server 30 according to the present embodiment. As shown in FIG. 3, the server 30 includes a communication unit 310, a learning data DB (Data Base) 320, and a control unit 330.

[0029] (1) Communications Unit 310 The communication unit 310 is a communication module for transmitting and receiving data to and from other devices via wired or wireless communication. For example, the communication unit 310 communicates with the sensor device 100 or the external DB 40.

[0030] (2) Learning Data DB320 The learning data DB320 stores learning data. The learning data is information including a feature vector and accident information. The accident information may be acquired from the sensor device 100 connected to the system 1, or may be acquired from the external DB40. The accident information is information including whether an accident has occurred, and if so, the details of the accident (geographical location, type of accident, date and time of occurrence, details of the occurrence, attributes of the assailant and victim, and identification information of the vehicle related to the accident). The external DB40 is a DB that accumulates accident information, for example, operated by a news organization, a transportation organization, a country, etc. Note that an accident in this specification refers to a so-called traffic accident such as a personal injury accident or a property damage accident.

[0031] (3) Control unit 330 The control unit 330 provides various functions of the server 30. The control unit 330 includes a memory control unit 332 and a learning unit 334. The control unit 330 may further include other components in addition to these components. That is, the control unit 330 may also perform operations other than those of these components. The operations of the memory control unit 332 and the learning unit 334 will be described in detail later.

[0032] <<3. Technical features>> Next, technical features of the system 1 according to this embodiment will be described. In the following, the technical features will be described focusing on one sensor device 100 mounted on a vehicle. The vehicle on which the sensor device 100 is mounted is also referred to as the subject vehicle, and other vehicles are also referred to as other vehicles. Sensor devices other than the sensor device 100 of interest are also referred to as other sensor devices.

[0033] (1) Accident probability prediction The sensor device 100 (for example, the prediction unit 152) predicts the accident probability of a vehicle (host vehicle) driven by a user (driver). In the following, unless otherwise specified, the driver refers to the driver of the host vehicle.

[0034] The sensor device 100 predicts the accident probability based on the feature quantities related to the driver or the vehicle. This makes it possible to predict the accident probability personalized for the driver or the vehicle. The sensor device 100 may also predict the accident probability based on the feature quantities related to other vehicles. This makes it possible to predict the accident probability personalized for the driver or the vehicle, taking into account the relationship with other vehicles. A feature quantity vector, which is a combination of one or more feature quantities, may be used to predict the accident probability. Specific examples of the feature quantity vector will be described in detail later.

[0035] The sensor device 100 predicts the accident probability using various methods. In the following, a prediction method using a logistic regression model will be described as an example. An example of the prediction formula is shown in the following formula (1).

[0036]

number

[0037] Where:

number

[0038]

number

[0039]

number

[0040] The server 30 (e.g., the memory control unit 332 and the learning data DB 320) accumulates the feature vector x transmitted from each vehicle (more precisely, the sensor device 100 mounted on each vehicle). Furthermore, the server 30 accumulates accident information transmitted from each vehicle or the external DB 40. Therefore, the server 30 holds a large number of (x, y) pairs. The server 30 (e.g., the learning unit 334) learns a prediction model (e.g., a parameter w) by using these pairs as learning data.

[0041] For example, the following describes the type of training data used to train a prediction model for predicting the probability of an accident after a predetermined time (e.g., after one minute). The training data includes a pair of feature vectors x and y=1 acquired when an accident occurs within the predetermined time (e.g., within one minute). The training data also includes a pair of feature vectors x and y=0 acquired when no accident occurs within the predetermined time (e.g., within one minute).

[0042] (2) Feature vector The sensor device 100 (e.g., the prediction unit 152) acquires a feature vector used to predict the accident probability. The feature vector may be generated in the sensor device 100 or in another device (e.g., the server 30 or another sensor device 100). As an example, a case where a feature vector is generated in the sensor device 100 will be described. For example, the sensor device 100 (e.g., the prediction unit 152) generates a feature vector based on sensor information detected by the detection unit 110 or information acquired by the communication unit 120 from another device (e.g., another sensor device 100 or the server 30, etc.). The symbol feature is converted into a numerical vector as a one-of-k feature vector.

[0043] The sensor device 100 (for example, the prediction unit 152) concatenates one or more acquired feature vectors. The sensor device 100 predicts the accident probability by inputting the concatenated feature vector into a prediction model.

[0044] The feature vector may be classified into, for example, a driving feature vector or an attribute feature vector. The driving feature vector is a feature vector based on sensor information detected in real time. Specifically, the driving feature vector may be a feature vector related to real-time driving for a predetermined period (e.g., 10 seconds). The driving feature vector is used to predict the accident probability, thereby making it possible to predict the accident probability according to the real-time situation. The attribute feature vector is a feature vector based on accumulated or registered information. Specifically, the attribute feature vector may be a feature vector related to the characteristics of the driver or the vehicle. The attribute feature vector is used to predict the accident probability, thereby making it possible to predict the accident probability according to the characteristics of the driver or the vehicle.

[0045] The feature vector can also be used to learn a prediction model of accident probability. For this purpose, the sensor device 100 (e.g., the communication unit 120) periodically transmits the sensor information or the feature vector to the server 30. At that time, the sensor device 100 transmits the sensor information or the feature vector in association with the identification information of the vehicle itself, the presence or absence of an accident, and, if an accident occurs, information indicating the details of the accident.

[0046] The sensor device 100 (e.g., the communication unit 120) may transmit only sensor information or feature vectors for a time period in which the predicted accident probability was equal to or greater than a threshold value. The sensor device 100 may also transmit only sensor information or feature vectors for a predetermined time period before and after the time when an accident or a precursor phenomenon, which will be described later, occurred. This makes it possible to reduce the amount of data transmission. Also, it is possible to prevent the prediction model from being overly adapted to learning data in which no accident or precursor phenomenon has occurred.

[0047] The driving feature quantity vector and the attribute feature quantity vector will be specifically described below.

[0048] (2.1) Driving feature vector An example of items of the feature vector that may be included in the driving feature vector is shown below. The driving feature vector is generated by concatenating one or more of the feature vectors shown below as examples.

[0049] (External object information) The external object information is a feature vector related to an object present outside the vehicle, detected by a camera, a radar, etc. The external object information may include, for example, the coordinates, size, speed and direction, acceleration and direction, reliability of the information, and detection time of the object.

[0050] (Vehicle movement information) The vehicle motion information is a feature vector that indicates the motion state of the host vehicle, detected by an acceleration sensor, etc. The vehicle motion information may include, for example, the speed, acceleration, and orientations of the host vehicle.

[0051] (Vehicle operation information) The vehicle operation information is a feature vector that indicates the state of operation of the vehicle by the driver. The vehicle operation information may include, for example, the following information: · Whether the turn signal is working or not ·Whether the wipers are working ·Angle of each mirror Whether each window is open or not, and the percentage of open windows · Whether the light is on or not Shift lever position - Degree of braking - Degree of accelerator pedal pressure Whether the clutch pedal is depressed or not

[0052] (Biometric information) The biometric information is a feature vector indicating the biometric information of a passenger (particularly, the driver). The biometric information may include information indicating, for example, blood pressure, heart rate, body temperature, blood sugar, and sweat rate. The biometric information may be detected by a biometric sensor provided in the vehicle. Alternatively, the biometric information may be detected by a biometric sensor provided in a mobile device carried by the passenger, and transmitted to the system 1 (for example, the sensor device 100 mounted on the vehicle in which the passenger is riding). The biometric information may be a statistical quantity, such as an average value of time-series changes in a measurement value.

[0053] (Biological status information) The biological condition information is a feature vector that indicates the state of the occupant, estimated based on the biological information. The biological condition information may include information indicating the concentration level, hunger level, degree of drowsiness, etc. This information may be estimated based on, for example, blood glucose level, blink frequency, degree of gaze blur, etc.

[0054] (In-car environment information) The in-vehicle environment information is a feature vector indicating the in-vehicle environment. The in-vehicle environment information may include, for example, information indicating the number of passengers. The information indicating the number of passengers may be recognized, for example, by a camera in the vehicle or a pressure sensor provided on the seat. The in-vehicle environment information may include, for example, information indicating temperature, humidity, and air pressure. The in-vehicle environment information may include, for example, information indicating the degree of fogging of the windows. The information indicating the degree of fogging of the windows may be recognized, for example, based on an image of the window captured by a camera in the vehicle. The in-vehicle environment information may include, for example, information indicating whether or not a conversation is taking place. The information indicating whether or not a conversation is taking place may be recognized, for example, based on a voice collected by a microphone in the vehicle. The in-vehicle environment information may include, for example, information indicating the volume of music. The information indicating the volume of music may be, for example, a setting value of a car stereo, or may be recognized based on a voice collected by a microphone in the vehicle. The in-vehicle environment information may be a statistic such as an average value of a time series change in a measurement value.

[0055] (Driver physical state information) The driver physical state information is a feature vector indicating the physical state of the driver. The driver physical state information may include, for example, information indicating the line of sight of the driver. The information indicating the line of sight of the driver may be, for example, information indicating the movement of the eyeballs over a predetermined period of time, the state of the pupils and their changes, and may be recognized by image recognition based on an image of the eyeballs. The driver physical state information may include, for example, information indicating the way the steering wheel is turned. The information indicating the way the steering wheel is turned may be, for example, information indicating the pattern of how the steering wheel is turned, or may be a statistic such as an average value of the time-series change in the angular velocity of the steering wheel. The driver physical state information may include, for example, information indicating the way the steering wheel is gripped. The information indicating the way the steering wheel is gripped may be, for example, information indicating whether the steering wheel is gripped with both hands or information indicating the gripping location, and may be recognized by image recognition or a sensor provided on the steering wheel. The driver physical state information may include, for example, information indicating the driver's posture. The information indicating the driver's posture may be, for example, information indicating the state of the seat (position and angle of the seat surface and backrest), and may be recognized by a sensor provided on the seat. The driver physical state information may be, for example, information indicating whether the driver is fastening a seat belt, and can be recognized by a sensor provided on a seat belt fastener.

[0056] (Weather information) The weather information is a feature vector that indicates the weather at the location where the vehicle is traveling. The weather information may include, for example, the following information: -Whether it's raining or snowing Temperature and humidity -Presence or absence of fog

[0057] (Road information) The road information is a feature vector that indicates the state of the road on which the vehicle is traveling. The road information may be recognized based on, for example, an image recognition result of an image captured by an external camera or map information. The road information may include, for example, the following information: · Whether it's a slope or not · On or off the highway · Bridge or not -Is the road paved?

[0058] (Accident information linked to maps) The accident information linked to the map is a feature vector indicating accidents that have occurred in the past, and includes information associated with location information of the accident location. For example, the feature vector of a certain intersection or curve is created and linked from each item of accident information that has occurred within a predetermined distance of the intersection or curve. The feature vector used when predicting the accident probability may be the sum of the feature vectors of intersections or curves that the vehicle is predicted to pass through within a predetermined time. The intersections or curves that the vehicle is predicted to pass through within a predetermined time are assumed to be intersections or curves that the vehicle is predicted to pass through if it continues traveling straight along the road and maintains the predicted speed of the vehicle.

[0059] (Cumulative driving time information) The cumulative driving time information is a feature vector that indicates a cumulative value of the driving time of the driver. For example, the cumulative value of the driving time may be a cumulative value from the last parking.

[0060] (2.2) Attribute feature vector An example of feature vector items that may be included in the attribute feature vector is shown below. The attribute feature vector is generated by concatenating one or more of the feature vectors shown below as examples.

[0061] (Driving history information) The driving history information is a feature vector indicating the driving history of the driver in a recent predetermined period (e.g., one year). Information indicating the driving history may include, for example, information such as the vehicle's speed, driving route, and braking timing. The driving history information may be a statistical quantity such as an average value of time-series changes in measured values. The driving history information may be linked to the driver or to the vehicle.

[0062] (Driver attribute information) The driver attribute information is a feature vector of the driver's attribute information, and may include, for example, the number of years of driving experience, the age, and the gender of the driver.

[0063] (Vehicle attribute information) The vehicle attribute information is a feature vector of the attribute information of the vehicle itself, and may include, for example, the vehicle type, model name, fuel efficiency, manufacturing year, and past repair history.

[0064] (Driving proficiency information) The driving proficiency information is a feature vector indicating the driving proficiency of the driver. The driving proficiency information is, for example, a five-point scale and can be self-reported by the driver. For drivers who do not self-report, the sensor device 100 (e.g., the control unit 150) can set the driving proficiency using a regression model. To this end, the sensor device 100 uses the driving proficiency as the objective variable, driving history information and driver attribute information as inputs, and creates a regression model by machine learning.

[0065] (Driver Habit Information) The driver's habit information is a feature vector indicating the driver's habit. The driver's habit information is, for example, the ID of the cluster to which the target driver belongs in the clustering result of the driving history information of multiple drivers. The clustering can be performed, for example, in the server 30. The clustering can be performed by predefining habit labels for some drivers, learning a classification model by machine learning as a classification problem with the driving history information as input for each habit label, and assigning the habit label to the driver. Specific habits include, for example, frequent lane changes, rough driving, etc.

[0066] (2.3) Supplementary Information When predicting the accident probability, for example, the sensor device 100 (e.g., the prediction unit 152) concatenates and uses the above-mentioned feature vectors. At that time, the feature vectors actually acquired when predicting the accident probability may be concatenated, or the feature vectors predicted to be acquired at the prediction target time (e.g., one minute later) may be concatenated.

[0067] Furthermore, the sensor device 100 may use not only the feature vector of the vehicle itself but also the feature vectors of other vehicles in the vicinity by concatenating them. For example, the sensor device 100 may concatenate the sum of the feature vectors of one or more other vehicles that are predicted to enter the vicinity (for example, within 100 m) after a predetermined time (for example, after 30 seconds) (if the vehicle is assumed to travel straight along the road and will pass by if the current speed is maintained) and that are predicted to enter the vicinity (for example, within 100 m) after a predetermined time (for example, after 30 seconds).

[0068] In addition, a pair of feature vectors may be treated as one feature vector. For example, a combination of driver habit information and accident information linked to a map can represent a specific driver habit that is likely to lead to a specific accident, and the accuracy of accident probability prediction can be improved.

[0069] (3) Learning The server 30 (for example, the learning unit 334) learns a prediction model used by the sensor device 100 when predicting the accident probability.

[0070] For example, the server 30 learns the parameter w in the above formula (1) based on a pair of the feature vector and information indicating the occurrence or non-occurrence of an accident stored in the learning data DB 320. Then, the server 30 (e.g., the communication unit 310) notifies information indicating the learning result (e.g., the parameter w) to all the sensor devices 100 connected to the system 1. Although a common prediction model is used among the multiple sensor devices 100, because the feature vector is personalized, personalized prediction of accident probability is possible.

[0071] The sensor device 100 (e.g., the storage unit 140) stores information indicating the learning result notified from the server 30. Then, at the time of prediction, the sensor device 100 (e.g., the prediction unit 152) uses the stored information to construct a prediction model. Therefore, the sensor device 100 does not need to communicate with the server 30 at the time of prediction.

[0072] The server 30 may learn a prediction model taking into account the occurrence probability of a precursor phenomenon of an accident (so-called near miss), such as sudden steering or sudden braking, in addition to the occurrence of an accident. As a result, the sensor device 100 (for example, the prediction unit 152) predicts the accident probability using a prediction model learned by taking into account the occurrence probability of a precursor phenomenon of an accident. Considering that the predicted result of the accident probability is notified (typically, a warning) to the driver, it is possible to prevent not only the occurrence of an accident but also the occurrence of the precursor phenomenon, and the safety of the vehicle can be further improved. Here, in order to generate a prediction model taking into account the precursor phenomenon in the prediction of the accident probability, the sensor device 100 transmits sensor information or a feature vector to the server 30 in association with the identification information of the vehicle, the presence or absence of a precursor phenomenon, and, if a precursor phenomenon occurs, information indicating the content of the precursor phenomenon.

[0073] Regarding the sudden steering, it may be determined that the steering wheel has been turned suddenly when the change in the rotational acceleration of the steering wheel and the change in the acceleration and direction of the vehicle are equal to or greater than a threshold value. Regarding the sudden braking, it may be determined that the braking has been turned suddenly when the change in the depression degree of the brake pedal and the change in the acceleration of the vehicle are equal to or greater than a threshold value. Other accident precursors may include ignoring traffic signals, stalling, running off the road, slipping, hydroplaning, and the like.

[0074] An example of a method for using precursor phenomena to predict the probability of an accident will be described below.

[0075] First example For example, the server 30 (e.g., the learning unit 334) learns a prediction model for the occurrence probability of a precursor phenomenon (e.g., a model similar to Equation (1)) in a similar manner to the prediction model for the accident probability. Next, the server 30 includes the occurrence probability of the precursor phenomenon in a feature vector to learn a prediction model related to the accident probability. The sensor device 100 (e.g., the prediction unit 152) first predicts the occurrence probability of the precursor phenomenon, and then predicts the accident probability by linking the occurrence probability of the precursor phenomenon to the feature vector.

[0076] Second Example For example, the server 30 (e.g., the learning unit 334) regards the prediction of the occurrence probability of a precursor phenomenon as a separate task from the prediction of the accident probability, and performs transfer learning.

number

[0077] The server 30 learns a prediction model for the occurrence probability of a precursor phenomenon using a logistic regression model, similar to the prediction model for the accident probability. The parameter obtained as a result of the learning is defined as w'. When learning the prediction model for the accident probability, the server 30 uses the following evaluation function.

[0078]

number

[0079] Here, n is the number of learning data, and α is an arbitrary constant. The first term above is the likelihood. The second term above enables the server 30 to learn a prediction model for the accident probability while taking into account the prediction model for the occurrence probability of a precursor phenomenon (so that the output of the prediction model does not deviate too much).

[0080] (4) User Interface The sensor device 100 outputs information related to the predicted result of the accident probability (for example, the output control unit 154 causes the output unit 130 to output). Below, an example of a UI when the sensor device 100 is mounted on a vehicle will be described with reference to Figs. 4 to 13. Figs. 4 to 13 are diagrams for explaining an example of a UI according to this embodiment.

[0081] (4.1) UI for accident probability itself The sensor device 100 may output information indicating the accident probability to the driver. This allows the driver to drive while grasping the fluctuation of the accident probability in real time. An example of the UI is shown in FIG. 4. In the example shown in FIG. 4, a speedometer 410 and an accident probability meter 420 are arranged on an instrument panel 400 (output unit 130). The accident probability meter 420 has a scale from 0 to 10 along the edge of a circle. The accident probability meter 420 outputs the accident probability by pointing a needle 421 with an axis at the center of the circle to a scale corresponding to a value obtained by multiplying the predicted accident probability (from 0 to 1) by 10. In the accident probability meter 420, an area 422 with a high value is highlighted.

[0082] The sensor device 100 may output the accident probability on a logarithmic scale. Considering that the frequency of accidents occurring is generally low, the accident probability is prevented from suddenly appearing to switch from a value close to 0% to a value close to 100% by expressing it logarithmically. An example of a UI is shown in FIG. 5. In the example shown in FIG. 5, a speedometer 410 and an accident probability meter 430 are arranged on an instrument panel 400 (output unit 130). In the accident probability meter 430, the accident probability is expressed on a logarithmic scale, as indicated by the scale on the edge of the circle. In the accident probability meter 430, a region 432 with a high value is also emphasized.

[0083] The sensor device 100 may output information indicating a time series transition of the accident probability. This allows the driver to drive while grasping the ups and downs of the accident probability from the past to the present in real time. Furthermore, the sensor device 100 may output a time series transition of the accident probability from the present to the future based on the prediction result of the future accident probability. An example of the UI is shown in FIG. 6. The UI shown in FIG. 6 shows a time series transition of the accident probability. The vertical axis is the accident probability, and the horizontal axis is time. The UI shown in FIG. 6 may be displayed on a display provided on, for example, an instrument panel or a car navigation system.

[0084] The sensor device 100 may output information based on the accumulated value of the accident probability. For example, the sensor device 100 first calculates the possible driving time by subtracting or adding a correction term based on biological condition information such as concentration, hunger, or drowsiness from a predetermined reference value such as 90 minutes from the start of driving. Then, the sensor device 100 calculates the accumulated value of the accident probability from the start of driving, and outputs the possible driving time or the possible driving distance calculated based on the possible driving time by reducing the larger the accumulated value. With such an output, the driver is encouraged to concentrate more on driving or take a break as the accumulated value increases, thereby further improving safety. An example of a UI is shown in FIG. 7. In the UI shown in FIG. 7, the possible driving distance calculated based on the remaining amount of fuel is displayed in the upper part 441 of the display 440 provided in the center of the speedometer 410, and the possible driving time described above is displayed in the lower part 442. Other possible display items based on the cumulative value of the accident probability include an item 443 showing remaining stamina (HP), an item 444 showing the distance that can be driven, and an item 445 showing the number of breaks required to reach the destination.

[0085] The sensor device 100 may output information on the accident probability on a map. For example, the sensor device 100 predicts the accident probability at an intersection or a curve using a prediction model constructed using only the attribute feature model, and outputs information indicating a point where the accident probability exceeds a threshold. In this case, it is desirable to predict the accident probability after updating the feature vector of the "accident information linked to the map" according to the intersection or curve on the planned driving route. By outputting information indicating a point where the accident probability exceeds a threshold, it is possible to alert the driver to a place where an accident is likely to occur, and safety can be improved. For the same purpose, information indicating a place where an accident has occurred in the past may be output. The sensor device 100 may also output information indicating a route with a low accident probability. In that case, it is possible to urge the driver to drive a safe route, and safety can be improved. For the same purpose, information indicating a route where no accidents have occurred in the past or where the number of accidents has been small may be output. Examples of the UI are shown in FIG. 8 and FIG. 9. In the UI shown in FIG. 8, the current position 450 of the vehicle, the planned driving route 451, and a point 452 where the accident probability exceeds a threshold are displayed on a map. 8 displays a message 453 warning that there is a point 452 where the accident probability exceeds a threshold. In the UI shown in FIG 9, a current position 450 of the vehicle, a planned driving route 451, a point 455 where many accidents have occurred in the past, and a route 454 with a low accident probability are displayed on a map. In addition, in the UI shown in FIG 9, a message 457 indicating the presence of the point 455 where many accidents have occurred in the past, and a message 456 indicating the presence of the route 454 with a low accident probability are displayed. In this way, safety can be further improved by highlighting information related to the accident probability on the map (for example, a point where the accident probability exceeds a threshold or a route with a low accident probability).

[0086] (4.2) UI for factors increasing accident probability The sensor device 100 may output information corresponding to the factor that increased the accident probability to the driver. For example, the sensor device 100 outputs information indicating the factor. This allows the driver to recognize the factor that increased the accident probability and to be alerted, thereby improving safety. The factor may be the feature amount itself, information estimated from the feature amount, or information estimated from the type of sensor corresponding to the feature amount. There may be a plurality of factors that increased the accident probability. In this case, the sensor device 100 may output information indicating a factor selected from one or more factors according to the contribution degree to the increase in the accident probability. Specifically, the sensor device 100 may output information indicating a factor that contributes greatly to the increase in the accident probability (for example, a factor whose contribution degree exceeds a threshold value, or a predetermined number of factors whose contribution degree is the highest). For example, when calculating the accident probability using the above formula (1), the feature amount itself that contributes greatly to the increase in the accident probability, information estimated from the feature amount, or information estimated from the type of sensor corresponding to the feature amount may be output. When a logistic regression model is used for prediction, the product of the value of the feature and the value of the parameter corresponding to the feature may be used as the contribution value. Such an output allows the driver to recognize factors with particularly high increase rates and to be alerted, thereby improving safety. An example of a UI is shown in FIG. 10. In the UI shown in FIG. 10, a message 460 indicating factors of an increase in the accident probability is displayed in the UI shown in FIG. 5. In addition to the message, for example, a predetermined icon for each factor may be displayed.

[0087] The sensor device 100 may output information instructing the driver to take an action to reduce the accident probability. Specifically, the sensor device 100 outputs a message to remove the factors that increased the accident probability described above. This makes it possible to prompt the driver to take a specific action, thereby improving safety. An example of the UI is shown in FIG. 11. In the UI shown in FIG. 11, a message 461 is displayed in the UI shown in FIG. 10, instructing the driver to take an action to reduce the accident probability.

[0088] (4.3) Other The sensor device 100 can provide various other UIs, examples of which are shown in FIGS.

[0089] 12, a voice message 471 corresponding to the driving situation is fed back as if it were coming from a character displayed on the display 470. For example, the sensor device 100 provides feedback to blame the driver for dangerous operations that may lead to accidents, such as sudden acceleration and sudden braking, and provides feedback to praise the driver for safe driving. This type of feedback allows the driver to drive safely as if playing a game.

[0090] In the UI shown in FIG. 13, information 482A-482C indicating the positional relationship between the vehicle 480 and the other vehicles 481A-481C and the percentage of fault in the case of an accident is output. For example, according to the information 482A, the percentage of fault in the case of an accident between the vehicle 480 and the other vehicle 481A is predicted to be 10 for the vehicle 480 and 0 for the other vehicle 481A. Also, according to the information 482B, the percentage of fault is predicted to be 0 for the vehicle 480 and 10 for the other vehicle 481A. Also, according to the information 482C, the percentage of fault is predicted to be 7 for the vehicle 480 and 3 for the other vehicle 481A. With such a UI, the driver can visually grasp which other vehicle he or she should be careful of. The UI shown in FIG. 13 can be displayed on, for example, an instrument panel or a display provided in a car navigation system.

[0091] <<4. Processing flow>> Next, an example of the flow of processing executed in the system 1 according to the present embodiment will be described with reference to FIGS.

[0092] (Predictive model learning process) 14 is a sequence diagram showing an example of the flow of a prediction model learning process executed in the system 1 according to this embodiment. The sensor device 100 and the server 30 are involved in this sequence.

[0093] As shown in FIG. 14, first, the sensor device 100 (e.g., the detection unit 110) detects sensor information (step S102). Next, the sensor device 100 (e.g., the prediction unit 152) generates a feature vector based on the detected sensor information (step S104). Next, the sensor device 100 (e.g., the communication unit 120) transmits the generated feature vector to the server 30 (step S106). Although omitted in FIG. 14, multiple sensor devices 100 connected to the system 1 transmit the feature vector to the server 30.

[0094] Next, the server 30 (e.g., the memory control unit 332 and the learning data DB 320) stores the feature vector received from one or more sensor devices 100, associating it with the vehicle identification information, whether an accident has occurred, and the details of the accident, as learning data (step S108). Then, the server 30 (e.g., the learning unit 334) learns a prediction model from the learning data (step S110). If the prediction model is the logistic regression model shown in formula (1), the server 30 learns the parameter w. Next, the server 30 (e.g., the communication unit 310) transmits information indicating the learning result (e.g., the parameter w) to all of the sensor devices 100 under its control (step S112).

[0095] Then, the sensor device 100 (for example, the prediction unit 152) updates the prediction model based on the received information indicating the learning result (step S114). For example, the prediction unit 152 updates the parameters of the prediction model to the received parameters.

[0096] This completes the process.

[0097] (Accident probability prediction processing) FIG. 15 is a flowchart showing an example of the flow of an accident probability prediction process executed in the sensor device 100 according to this embodiment.

[0098] As shown in FIG. 15, first, the sensor device 100 (for example, the detection unit 110) detects sensor information (step S202). Next, the sensor device 100 (for example, the prediction unit 152) generates a feature vector based on the detected sensor information (step S204). The generated feature vector may include not only the feature vector of the driver of the vehicle or the vehicle itself, but also the feature vector of the driver of another vehicle or the other vehicle. Next, the sensor device 100 (for example, the prediction unit 152) inputs the feature vector into a prediction model to predict the accident probability (step S206). Then, the sensor device 100 (for example, the output control unit 154) performs output processing (step S208). For example, the output control unit 154 causes the output unit 130 to output a UI related to the accident probability itself or a UI related to a factor of increasing the accident probability.

[0099] This completes the process.

[0100] (Feature vector generation process) 16 is a flowchart showing an example of the flow of the feature vector generation process executed in the sensor device 100 according to this embodiment. This flow corresponds to the processes in steps S104 and S204 described above.

[0101] As shown in Fig. 16, first, the sensor device 100 (e.g., the prediction unit 152) generates a driving feature vector based on the latest sensor information (step S302). Next, the sensor device 100 (e.g., the prediction unit 152) updates the attribute feature vector based on the driving feature vector (step S304). Then, the sensor device 100 (e.g., the prediction unit 152) concatenates the driving feature vector and the attribute feature vector (step S306).

[0102] This completes the process.

[0103] Next, as an example of the process of generating individual feature vectors, a process of generating a feature vector related to accident information linked to a map and a process of generating a feature vector related to the occurrence probability of a precursor phenomenon will be described.

[0104] FIG. 17 is a flowchart showing an example of the flow of a feature quantity vector generation process for accident information linked to a map, which is executed in the sensor device 100 according to this embodiment.

[0105] As shown in FIG. 17, first, the sensor device 100 (e.g., the prediction unit 152) identifies an intersection or a curve that is predicted to be passed within a predetermined period in the future (step S402). Next, the sensor device 100 (e.g., the prediction unit 152) acquires accident information associated with the identified intersection or curve from the server 30 (step S404). Next, the sensor device 100 (e.g., the prediction unit 152) generates a feature vector for each of the acquired accident information (step S406). Then, the sensor device 100 (e.g., the prediction unit 152) takes the sum of one or more generated feature vectors, and concatenates the sum to a driving feature vector (step S408).

[0106] This completes the process.

[0107] FIG. 18 is a flowchart showing an example of the flow of a feature quantity vector generation process relating to the occurrence probability of a precursor phenomenon, which is executed in the sensor device 100 according to this embodiment.

[0108] As shown in FIG. 18, first, the sensor device 100 (e.g., the prediction unit 152) predicts the occurrence probability of the precursor phenomenon using a prediction model (step S502). This prediction model is a prediction model for the occurrence probability of the precursor phenomenon. Next, the sensor device 100 (the prediction unit 152) links the occurrence probability of the precursor phenomenon to the driving feature vector (step S504).

[0109] This completes the process.

[0110] <<5. Hardware configuration example>> The technology according to the present disclosure can be applied to various products. For example, the sensor device 100 may be realized as a device mounted on any type of vehicle, such as an automobile, an electric vehicle, a hybrid electric vehicle, or a motorcycle. In addition, at least some of the components of the sensor device 100 may be realized in a module (for example, an integrated circuit module configured on a single die) for the device mounted on the vehicle.

[0111] <5.1. Example of vehicle control system configuration> 19 is a block diagram showing an example of a schematic configuration of a vehicle control system 900 to which the technology according to the present disclosure can be applied. The vehicle control system 900 includes an electronic control unit 902, a storage device 904, an input device 906, an outside sensor 908, a vehicle state sensor 910, an occupant sensor 912, a communication IF 914, an output device 916, a power generating device 918, a braking device 920, a steering device 922, and a lamp operating device 924.

[0112] The electronic control unit 902 functions as an arithmetic processing device and a control device, and controls the overall operation in the vehicle control system 900 according to various programs. The electronic control unit 902 may be formed as an ECU (Electronic Control Unit) together with a storage device 904 described later. A plurality of ECUs (i.e., the electronic control unit 902 and the storage device 904) may be included in the vehicle control system 900. For example, an ECU for controlling each of the various sensors or the various drive systems may be provided, and an ECU for cooperatively controlling the plurality of ECUs may be further provided. These plurality of ECUs are connected to each other via an in-vehicle communication network conforming to any standard such as CAN (Controller Area Network), LIN (Local Interconnect Network), LAN (Local Area Network), or Flexray. The electronic control unit 902 may form, for example, a control unit 150 shown in FIG. 2.

[0113] The storage device 904 is a data storage device formed as an example of a storage unit of the vehicle control system 900. The storage device 904 is realized by, for example, a magnetic storage device such as an HDD, a semiconductor storage device, an optical storage device, or a magneto-optical storage device. The storage device 904 may include a storage medium, a recording device for recording data in the storage medium, a reading device for reading data from the storage medium, and a deleting device for deleting data recorded in the storage medium. This storage device 904 stores programs and various data executed by the electronic control unit 902, and various data acquired from the outside. The storage device 904 may form, for example, the storage unit 140 shown in FIG. 2.

[0114] The input device 906 is realized by a device into which information is input by a passenger (driver or passenger), such as a mouse, a keyboard, a touch panel, a button, a microphone, a switch, and a lever. The input device 906 may be, for example, a remote control device using infrared or other radio waves, or an external connection device such as a mobile phone or a PDA that supports the operation of the vehicle control system 900. The input device 906 may be, for example, a camera, in which case the passenger can input information by gestures. The input device 906 may include, for example, an input control circuit that generates an input signal based on information input by a user using the above-mentioned input means and outputs the signal to the electronic control unit 902. The passenger can input various data to the vehicle control system 900 or instruct a processing operation by operating the input device 906. The input device 906 may form, for example, the detection unit 110 shown in FIG. 2.

[0115] The outside sensor 908 is realized by a sensor that detects information outside the vehicle. For example, the outside sensor 908 may include a sonar device, a radar device, a LIDAR (Light Detection and Ranging, Laser Imaging Detection and Ranging) device, a camera, a stereo camera, a ToF (Time Of Flight) camera, an infrared sensor, an environmental sensor, a microphone, etc. The outside sensor 908 may form, for example, the detection unit 110 shown in FIG. 2.

[0116] The vehicle state sensor 910 is realized by a sensor that detects information about the vehicle state. For example, the vehicle state sensor 910 may include a sensor that detects an operation by a driver, such as an accelerator opening, a brake pedal pressure, or a steering angle. The vehicle state sensor 910 may also include a sensor that detects a state of a power source, such as the rotation speed or torque of an internal combustion engine or a motor. The vehicle state sensor 910 may also include a sensor for detecting information about the movement of the vehicle, such as a gyro sensor or an acceleration sensor. The vehicle state sensor 910 may also include a GNSS module that receives a GNSS signal from a GNSS (Global Navigation Satellite System) satellite (for example, a GPS signal from a GPS (Global Positioning System) satellite) to measure position information including the latitude, longitude, and altitude of the device. Regarding the position information, the vehicle state sensor 910 may detect the position by transmission and reception with Wi-Fi (registered trademark), a mobile phone, a PHS, a smartphone, or the like, or by short-range communication. The vehicle state sensor 910 may form, for example, a detection unit 110 shown in FIG. 2.

[0117] The passenger sensor 912 is realized by a sensor that detects information about the passenger. For example, the passenger sensor 912 may include a camera, a microphone, and an environmental sensor provided in the vehicle. The passenger sensor 912 may also include a biosensor that detects bioinformation of the passenger. The biosensor is provided, for example, on the seat or steering wheel, and can detect bioinformation of the passenger sitting in the seat or the driver holding the steering wheel. The passenger sensor 912 may form, for example, the detection unit 110 shown in FIG. 2.

[0118] Each of the various sensors, such as the vehicle exterior sensor 908, the vehicle state sensor 910, and the passenger sensor 912, outputs information indicating the detection result to the electronic control unit 902. The various sensors may set the sensing range or accuracy based on the control of the electronic control unit 902. The various sensors may also include a recognition module that performs recognition processing based on raw data, such as processing for recognizing the running position of the vehicle on the road based on the position of a white line included in a captured image.

[0119] The communication IF 914 is a communication interface that mediates communication between the vehicle control system 900 and other devices. The communication IF 914 may include, for example, a V2X communication module. Note that V2X communication is a concept that includes vehicle-to-vehicle (V2X) communication and vehicle-to-infrastructure (V2X) communication. In addition, the communication IF 914 may include a communication module for a wireless LAN (Local Area Network), Wi-Fi (registered trademark), 3G, LTE (Long Term Evolution), Bluetooth (registered trademark), NFC (Near Field Communication), or WUSB (Wireless USB). The communication IF 914 can transmit and receive signals, for example, between the Internet or a communication device outside the vehicle in accordance with a predetermined protocol such as TCP / IP. The communication IF 914 may form, for example, the communication unit 120 shown in FIG. 2.

[0120] The output device 916 is realized by a device capable of visually or audibly notifying the passenger of the acquired information. Such devices include display devices such as an instrument panel, a head-up display, a projector, or a lamp, and audio output devices such as a speaker or a headphone. Specifically, the display device visually displays the results obtained by various processes performed by the vehicle control system 900 in various formats such as text, images, tables, and graphs. At that time, virtual objects such as AR (Augmented Reality) objects may be displayed. On the other hand, the audio output device converts audio signals consisting of reproduced audio data, acoustic data, and the like into analog signals and audibly outputs them. The display device and the audio output device may form, for example, the output unit 130 shown in FIG. 2.

[0121] The power generating device 918 is a device for generating a driving force for the vehicle. The power generating device 918 may be realized by, for example, an internal combustion engine. In this case, the power generating device 918 performs start control, stop control, throttle valve opening control, fuel injection control, EGR (Exhaust Gas Recirculation) control, and the like, based on a control command from the electronic control unit 902. The power generating device 918 may also be realized by, for example, a motor, an inverter, and a battery. In this case, the power generating device 918 can perform a motor operation (so-called power running) in which power is supplied from the battery to the motor via the inverter and a positive torque is output, and a regenerative operation in which the motor absorbs torque to generate power and charges the battery via the inverter, based on a control command from the electronic control unit 902.

[0122] The braking device 920 is a device for applying a braking force to the vehicle or for slowing or stopping the vehicle. The braking device 920 may include, for example, brakes installed on each wheel, and brake pipes or electric circuits for transmitting the pressure of the brake pedal to the brakes. The braking device 920 may also include a control device for activating a skid or side slip prevention mechanism by brake control such as an ABS (Antilock Brake System) or an ESC (Electronic Stability Control).

[0123] The steering wheel 922 is a device for controlling the traveling direction (steering angle) of the vehicle. The steering wheel 922 may include, for example, a steering wheel, a steering shaft, a steering gear, and a tie rod. The steering wheel 922 may also include a power steering for assisting the driver in steering. Furthermore, the steering wheel 922 may also include a power source such as a motor for realizing automatic steering.

[0124] The lamp operating device 924 is a device that operates various lamps such as headlights, blinkers, width lights, fog lights, stop lamps, etc. The lamp operating device 924 controls, for example, the blinking, light amount, or irradiation direction of the lamps.

[0125] The power generating device 918, the braking device 920, the steering 922, and the lamp operating device 924 may operate based on manual operation by the driver, or based on automatic operation by the electronic control unit 902.

[0126] <5.2. Example of information processing device configuration> Fig. 20 is a block diagram showing an example of a hardware configuration of an information processing device according to this embodiment. The information processing device 1000 shown in Fig. 20 can realize, for example, the server 30 shown in Fig. 3. Information processing by the server 30 according to this embodiment is realized by cooperation between software and hardware described below.

[0127] 20, the information processing device 1000 includes a CPU (Central Processing Unit) 1001, a ROM (Read Only Memory) 1002, a RAM (Random Access Memory) 1003, and a host bus 1004a. The information processing device 1000 also includes a bridge 1004, an external bus 1004b, an interface 1005, an input device 1006, an output device 1007, a storage device 1008, a drive 1009, a connection port 1011, and a communication device 1013. The information processing device 1000 may include a processing circuit such as a DSP or an ASIC instead of or in addition to the CPU 1001.

[0128] The CPU 1001 functions as an arithmetic processing device and a control device, and controls the overall operation of the information processing device 1000 in accordance with various programs. The CPU 1001 may also be a microprocessor. The ROM 1002 stores programs and arithmetic parameters used by the CPU 1001. The RAM 1003 temporarily stores programs used in the execution of the CPU 1001 and parameters that change appropriately during the execution. The CPU 1001 may form, for example, a control unit 330 shown in FIG. 3.

[0129] The CPU 1001, ROM 1002, and RAM 1003 are connected to one another by a host bus 1004a including a CPU bus, etc. The host bus 1004a is connected to an external bus 1004b such as a PCI (Peripheral Component Interconnect / Interface) bus via a bridge 1004. Note that the host bus 1004a, bridge 1004, and external bus 1004b do not necessarily need to be configured separately, and these functions may be implemented on a single bus.

[0130] The input device 1006 is realized by a device into which information is input by a user, such as a mouse, a keyboard, a touch panel, a button, a microphone, a switch, and a lever. The input device 1006 may be, for example, a remote control device using infrared or other radio waves, or an externally connected device such as a mobile phone or a PDA that supports the operation of the information processing device 1000. The input device 1006 may further include, for example, an input control circuit that generates an input signal based on information input by a user using the above-mentioned input means and outputs the signal to the CPU 1001. A user of the information processing device 1000 can input various data to the information processing device 1000 and instruct the information processing device 1000 to perform processing operations by operating the input device 1006.

[0131] The output device 1007 is formed of a device capable of visually or aurally notifying the user of acquired information. Such devices include display devices such as CRT display devices, liquid crystal display devices, plasma display devices, EL display devices, laser projectors, LED projectors, and lamps, audio output devices such as speakers and headphones, and printer devices. The output device 1007 outputs, for example, results obtained by various processes performed by the information processing device 1000. Specifically, the display device visually displays the results obtained by various processes performed by the information processing device 1000 in various formats such as text, images, tables, and graphs. On the other hand, the audio output device converts audio signals consisting of reproduced voice data, acoustic data, and the like, into analog signals and aurally outputs them.

[0132] The storage device 1008 is a data storage device formed as an example of a storage unit of the information processing device 1000. The storage device 1008 is realized by, for example, a magnetic storage device such as an HDD, a semiconductor storage device, an optical storage device, or a magneto-optical storage device. The storage device 1008 may include a storage medium, a recording device for recording data in the storage medium, a reading device for reading data from the storage medium, and a deleting device for deleting data recorded in the storage medium. The storage device 1008 stores programs and various data executed by the CPU 1001, various data acquired from the outside, and the like. The storage device 1008 may form, for example, a learning data DB 320 shown in FIG. 3.

[0133] The drive 1009 is a reader / writer for a storage medium, and is built into or externally attached to the information processing device 1000. The drive 1009 reads information recorded on a removable storage medium, such as an attached magnetic disk, optical disk, magneto-optical disk, or semiconductor memory, and outputs the information to the RAM 1003. The drive 1009 can also write information to the removable storage medium.

[0134] The connection port 1011 is an interface connected to an external device, and is a connection port for connecting to an external device capable of transmitting data via, for example, a USB (Universal Serial Bus) or the like.

[0135] The communication device 1013 is, for example, a communication interface formed by a communication device for connecting to the network 1020. The communication device 1013 is, for example, a communication card for a wired or wireless LAN (Local Area Network), LTE (Long Term Evolution), Bluetooth (registered trademark), or WUSB (Wireless USB). The communication device 1013 may also be a router for optical communication, a router for ADSL (Asymmetric Digital Subscriber Line), or a modem for various communications. The communication device 1013 can transmit and receive signals, for example, between the Internet and other communication devices in accordance with a predetermined protocol such as TCP / IP. The communication device 1013 can form, for example, a communication unit 310 shown in FIG. 3.

[0136] The network 1020 is a wired or wireless transmission path for information transmitted from devices connected to the network 1020. For example, the network 1020 may include public line networks such as the Internet, telephone lines, and satellite communication networks, as well as various LANs (Local Area Networks) including Ethernet (registered trademark), and WANs (Wide Area Networks). The network 1020 may also include a dedicated line network such as an IP-VPN (Internet Protocol-Virtual Private Network).

[0137] <5.3. Supplementary Information> An example of a hardware configuration capable of realizing the functions of the sensor device 100 or the server 30 according to the present embodiment has been described above. Each of the above components may be realized using general-purpose members, or may be realized by hardware specialized for the function of each component. Therefore, the hardware configuration to be used can be changed as appropriate depending on the technical level at the time of implementing the present embodiment.

[0138] It is possible to create a computer program for implementing each function of the sensor device 100 or the server 30 according to the present embodiment as described above, and to install it in an ECU or the like. It is also possible to provide a computer-readable recording medium in which such a computer program is stored. The recording medium is, for example, a magnetic disk, an optical disk, a magneto-optical disk, a flash memory, or the like. The above computer program may also be distributed, for example, via a network, without using a recording medium.

[0139] <<6. Summary>> An embodiment of the present disclosure has been described above in detail with reference to Figs. 1 to 19. As described above, the sensor device 100 according to this embodiment predicts the accident probability of a vehicle driven by a user, and outputs information to the user according to factors that increase the predicted accident probability. This clearly indicates what specific accident avoidance actions should be taken when the accident probability is high, so that the user can easily take the accident avoidance action and increase safety.

[0140] In addition, the sensor device 100 according to the present embodiment predicts the accident probability using various feature vectors that may affect the occurrence of an accident. For example, the sensor device 100 predicts the accident probability using various feature vectors that may affect the occurrence of an accident, such as forgetting to turn on the blinker (vehicle operation information), decreased concentration (biological condition information), poor physical condition (biological information), fogged windows (in-vehicle environment information), rain or night (weather information), not looking around (driver physical condition information), steering with one hand (driver physical condition information), intersections where accidents are likely to occur (accident information linked to a map), rough driving (driver habit information), and a reckless car ahead of the intersection (driving proficiency information or accident probability of other nearby vehicles). This can improve the prediction accuracy of the accident probability.

[0141] Although the preferred embodiment of the present disclosure has been described in detail above with reference to the attached drawings, the technical scope of the present disclosure is not limited to such examples. It is clear that a person having ordinary knowledge in the technical field of the present disclosure can conceive of various modified or amended examples within the scope of the technical ideas described in the claims, and it is understood that these also naturally belong to the technical scope of the present disclosure.

[0142] For example, in the above embodiment, an example in which the sensor device 100 is mounted on a vehicle has been mainly described, but the present technology is not limited to such an example. For example, the sensor device 100 can be mounted on any moving object such as an airplane, a bicycle, or a motorcycle. The sensor device 100 may be realized as a user device such as a smartphone or a head mounted display (HMD). The sensor device 100 may be realized as a device installed in the environment such as a traffic light, a surveillance camera, a digital signage, or an electronic bulletin board. In this case, the sensor device 100 outputs information to a driver of a vehicle traveling in the vicinity, for example.

[0143] In addition, each device described in this specification may be realized as a single device, or some or all of them may be realized as separate devices. For example, among the functional configuration example of the sensor device 100 shown in Fig. 2, the storage unit 140 and the control unit 150 may be provided in a device such as a server connected to the detection unit 110, the communication unit 120, and the output unit 130 via a network or the like.

[0144] In addition, the processes described in this specification using flowcharts and sequence diagrams do not necessarily have to be performed in the order shown in the drawings. Some processing steps may be performed in parallel. In addition, additional processing steps may be employed, and some processing steps may be omitted.

[0145] In addition, the effects described in this specification are merely descriptive or exemplary and are not limiting. In other words, the technology according to the present disclosure may achieve other effects that are apparent to a person skilled in the art from the description of this specification, in addition to or in place of the above effects.

[0146] Note that the following configurations also fall within the technical scope of the present disclosure. (1) A prediction unit that predicts an accident probability of a vehicle driven by a user; an output control unit that outputs information corresponding to a factor that has increased the accident probability predicted by the prediction unit to the user; An information processing device comprising: (2) The information processing device according to (1), wherein the output control unit outputs information indicating the cause. (3) The information processing device according to (2), wherein the output control unit outputs information indicating a factor selected from one or more of the factors according to its contribution to an increase in the accident probability. (4) The information processing device according to any one of (1) to (3), wherein the output control unit outputs information instructing the user on an action to be taken in order to reduce the accident probability. (5) The information processing device according to any one of (1) to (4), wherein the output control unit outputs information indicating the accident probability. (6) The information processing device according to (5), wherein the output control unit outputs the accident probability on a logarithmic scale. (7) The information processing device according to (5) or (6), wherein the output control unit outputs information indicating a time series transition of the accident probability. (8) The information processing device according to any one of (5) to (7), wherein the output control unit outputs information based on a cumulative value of the accident probability. (9) The information processing device according to any one of (5) to (8), wherein the output control unit outputs information about the accident probability on a map. (10) The information processing device according to any one of (1) to (9), wherein the prediction unit predicts the accident probability using a prediction model that is trained taking into account a probability of occurrence of a precursor phenomenon of an accident. (11) The information processing device according to (10), wherein the precursor phenomenon is sudden braking or sudden steering. (12) The information processing device according to any one of (1) to (11), wherein the prediction unit predicts the accident probability based on a feature amount related to the user or the vehicle. (13) The information processing device according to (12), wherein the feature amount is information detected in real time. (14) The information processing device according to (12), wherein the feature is accumulated or registered information. (15) The information processing device according to any one of (12) to (14), wherein the prediction unit predicts the accident probability based on a feature amount related to another vehicle. (16) Predicting an accident probability of a vehicle driven by a user; outputting information corresponding to a factor that has increased the predicted accident probability to the user via an output device; An information processing method comprising: (17) Computer, A prediction unit that predicts an accident probability of a vehicle driven by a user; an output control unit that outputs information corresponding to a factor that has increased the accident probability predicted by the prediction unit to the user; A program to function as a [Explanation of symbols]

[0147] 1 System 100 Sensor device 110 Detection unit 120 Communications Department 130 Output section 140 Storage section 150 Control section 152 Prediction Department 154 Output control section 30 Servers 310 Communications Department 320 Learning Data DB 330 Control Unit 332 Memory control unit 334 Learning Department 40 External DB

Claims

1. A generation unit that generates a feature vector based on sensor information related to a vehicle or a driver; an acquisition unit that acquires information about locations where accidents are likely to occur, the information being obtained by inputting the feature vector into a prediction model; an output control unit that controls output of the location and a warning for the location on a map; An information processing system comprising:

2. The location includes a location where the accident probability is equal to or exceeds a threshold, or a location where an accident has occurred in the past. The information processing system according to claim 1 .

3. The output control unit further controls the output of routes with a low accident probability on a map.

3. The information processing system according to claim 1 or 2.

4. The output control unit further controls output of the current position of the vehicle on a map.

4. An information processing system according to claim 1.

5. The acquisition unit acquires information regarding factors that increase the accident probability, The output control unit controls output of a predetermined warning for each of the factors.

5. An information processing system according to claim 1.

6. The factor is a feature based on the sensor information, information estimated from the feature, or information estimated from the type of sensor.

6. The information processing system according to claim 5.

7. The feature vector includes at least one of a driving feature vector based on the sensor information detected in real time and an attribute feature vector related to the characteristics of the driver or the vehicle.

7. An information processing system according to claim 1.

8. The prediction model is trained using training data including the feature vector and accident information.

8. An information processing system according to claim 1.

9. the acquisition unit acquires prediction information obtained by inputting the feature vector into the prediction model, the prediction information including information about the location; The output control unit outputs the prediction information on a map.

9. An information processing system according to claim 1.

10. The information processing system generating a feature vector based on sensor information relating to a vehicle or a driver; Inputting the feature vector into a prediction model to obtain information about locations where accidents are likely to occur, Controlling the output of said location and warnings for said location on a map Information processing methods.

11. On the computer, generating a feature vector based on sensor information relating to a vehicle or a driver; Inputting the feature vector into a prediction model to obtain information about locations where accidents are likely to occur, Controlling the output of said location and warnings for said location on a map A program for executing a process.

Images