Information processing device, information processing method, and computer program

Abstract

To specify more detailed driving risk characteristics of drivers.SOLUTION: An information processing device includes a data acquisition part and a classification processing part. The data acquisition part acquires acceleration data showing a combination of lateral acceleration and longitudinal acceleration of a movable body at each timing during driving of the movable body for a plurality of drivers. The classification processing part classifies a plurality of dangerous driving actions by the drivers into a plurality of dangerous driving action patterns set on the basis of the combination of the lateral acceleration and the longitudinal acceleration by executing clustering using the acceleration data for each of the plurality of drivers, and outputs a classification result.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The technology disclosed in this specification relates to information processing for identifying driving risk characteristics of a driver of a mobile object. [Background technology]

[0002] Identifying the driving risk characteristics of drivers of vehicles such as automobiles is useful in various situations. For example, by providing feedback on the identified driving risk characteristics to the driver or by using the identified driving risk characteristics for insurance, the driver can be encouraged to improve their driving behavior.

[0003] BACKGROUND ART Conventionally, there is known a technique for evaluating the driving risk characteristics of a vehicle driver based on acceleration data indicating a combination of the lateral acceleration and longitudinal acceleration of the vehicle (see, for example, Non-Patent Document 1). [Prior art documents] [Non-patent literature]

[0004] [Non-Patent Document 1] Johan W. Joubert and two others, "Combining accelerometer data and contextual variables to evaluate the risk of driver behavior," Transportation Research Part F, Elsevier, July 5, 2016, Issue 41, pp. 80-96 Summary of the Invention [Problem to be solved by the invention]

[0005] Conventional techniques have room for improvement in identifying more detailed driving risk characteristics of drivers.

[0006] This specification discloses a technique that can solve the above-mentioned problems. [Means for solving the problem]

[0007] The technology disclosed in this specification can be realized, for example, in the following forms.

[0008] (1) An information processing device disclosed in this specification includes a data acquisition unit and a classification processing unit. The data acquisition unit acquires acceleration data for a plurality of drivers, the acceleration data indicating a combination of lateral acceleration and vertical acceleration of the vehicle at each timing when the vehicle is being driven. The classification processing unit performs clustering using the acceleration data for each of the plurality of drivers to classify a plurality of risky driving behaviors of the drivers into a plurality of risky driving behavior patterns set based on the combinations of the lateral acceleration and the vertical acceleration, and outputs the classification results.

[0009] This information processing device can output the results of classifying multiple risky driving behaviors by a driver into multiple risky driving behavior patterns set based on a combination of lateral acceleration and longitudinal acceleration, thereby making it possible to identify the driver's driving risk characteristics in more detail.

[0010] (2) In the information processing device, the classification result may include information indicating the proportion of the risky driving behaviors that belong to each of the risky driving behavior patterns. With this configuration, it is possible to identify the driver's driving risk characteristics in more detail in a more easily understandable manner.

[0011] (3) In the information processing device, the classification processing unit may be configured to perform the clustering using a Gaussian mixture distribution model. With this configuration, it is possible to accurately identify more detailed driving risk characteristics of a driver.

[0012] (4) In the information processing device, the acceleration data may include location information indicating the location of the moving object at each timing, and the information processing device may further include a spatiotemporal information processing unit that creates spatiotemporal information indicating the time and location of the occurrence of the risky driving behavior and outputs the spatiotemporal information. With this configuration, it is possible to identify times and places (hot spots) where risky driving behavior frequently occurs for each driver, thereby clarifying implicit factors behind the driver's risky driving behavior and enabling a more fair comparison of drivers' driving performance.

[0013] (5) Another information processing device disclosed in this specification includes a data acquisition unit and a classification processing unit. The data acquisition unit acquires acceleration data for a plurality of drivers, the acceleration data indicating a combination of lateral acceleration and longitudinal acceleration of the vehicle at each timing while the vehicle is being driven. The classification processing unit performs clustering using the acceleration data to classify the plurality of drivers into a plurality of driving behavior groups having different driving risk levels, and outputs the classification results.

[0014] According to this information processing device, it is possible to output the results of classifying a plurality of drivers into a plurality of driving behavior groups with different driving risk levels, and to identify the driving risk characteristics of the drivers in more detail.

[0015] (6) In the information processing device, the classification processing unit may be configured to calculate a first parameter indicating a proportion of driving operations that correspond to a predetermined dangerous acceleration operation and a second parameter indicating instability of driving behavior based on the acceleration data, and to perform the clustering using the first parameter and the second parameter. With this configuration, it is possible to accurately identify more detailed driving risk characteristics of the driver.

[0016] (7) In the information processing device, the classification processing unit may be configured to perform the clustering using a hierarchical clustering algorithm. With this configuration, it is possible to accurately identify more detailed driving risk characteristics of a driver.

[0017] (8) In the information processing device, the classification processing unit may further perform clustering using the acceleration data for each of the plurality of drivers to classify the plurality of risky driving behaviors of the drivers into a plurality of risky driving behavior patterns set based on combinations of the lateral acceleration and the longitudinal acceleration, and output the classification results. With this configuration, it is possible to identify more detailed driving risk characteristics of the drivers.

[0018] The technology disclosed in this specification can be realized in various forms, such as an information processing device, an information processing method, a computer program that realizes those methods, a non-transitory recording medium on which that computer program is recorded, etc. [Brief explanation of the drawings]

[0019] [Figure 1] FIG. 10 is an explanatory diagram showing an example of a driving risk profile RP generated in this embodiment. [Figure 2] FIG. 1 is an explanatory diagram showing a schematic configuration of an information processing device 100. [Figure 3] Flowchart showing driver evaluation processing [Figure 4] An explanatory diagram showing an example of GG Diagram 20 [Figure 5] An explanatory diagram showing an example of a safe driving area SA [Figure 6] FIG. 10 is an explanatory diagram showing an example of the result of the first clustering process C1. [Figure 7] FIG. 1 is an explanatory diagram showing an example of how spatiotemporal information is used; [Figure 8] FIG. 1 is an explanatory diagram showing an example of how spatiotemporal information is used; [Figure 9] An explanatory diagram showing another example of a driving risk profile RP. DETAILED DESCRIPTION OF THE INVENTION

[0020] A. Implementation: A-1. Overview of the Driving Risk Profile RP: First, an overview of the driving risk profile RP generated in this embodiment will be described. Fig. 1 is an explanatory diagram showing an example of the driving risk profile RP generated in this embodiment.

[0021] As shown in FIG. 1, the driving risk profile RP is information indicating the driving risk characteristics of each driver identified by a driver ID. More specifically, the driving risk profile RP includes information specifying to which of a plurality of driving behavior groups DG each driver belongs, each having different driving risk levels. In the example of FIG. 1, three driving behavior groups DG (DG1: high-risk driver group, DG2: medium-risk driver group, and DG3: low-risk driver group) are set, and each driver is classified into one of the three driving behavior groups DG. For example, the driving risk profile RP indicates that five drivers with driver IDs = 2, 15, 36, 44, and 226 belong to DG1: high-risk driver group.

[0022] The driving risk profile RP also includes information indicating the type of risky driving behavior each driver is prone to. Specifically, the driving risk profile RP classifies each driver's risky driving behaviors into multiple risky driving behavior patterns DP and includes information indicating the proportion of risky driving behaviors belonging to each of the multiple risky driving behavior patterns DP. In this embodiment, as shown in Table 1 below, eight risky driving behavior patterns DP (DP1: dangerous acceleration, DP2: dangerous deceleration, DP3: dangerous left cornering, DP4: dangerous right cornering, DP5: dangerous left turn with acceleration, DP6: dangerous left turn with deceleration, DP7: dangerous right turn with acceleration, and DP8: dangerous right turn with deceleration) are set, and the driver's risky driving behaviors are classified into one of the eight risky driving behavior patterns DP. For example, the driving risk profile RP indicates that the proportion of risky driving behaviors belonging to risky driving behavior pattern DP5 (risky left turn with acceleration) for the driver with driver ID=226 is approximately 80%, indicating that the majority of risky driving behaviors are dangerous left turns with acceleration. In addition, the driving risk profile RP indicates that the driver with driver ID=44 tends to engage in risky driving behavior when accelerating, since all of his risky driving behaviors belong to one of the risky driving behavior patterns DP (DP1, DP5, DP7) that involve acceleration.

[0023] In this way, by referring to the driving risk profile RP of this embodiment, it is possible to understand what type of driver each driver is in terms of driving risk and what types of risky driving behavior each driver tends to engage in. Therefore, by referring to the driving risk profile RP of this embodiment, it is possible to identify more detailed driving risk characteristics of the driver, and as a result, the usefulness of the driving risk profile RP can be improved.

[0024] The example driving risk profile RP described in this specification was created using driving history data (total driving distance: 830,338 km) of 71 drivers (aged 54 to 87, average age 72.61) collected from February 2015 to 2019.

[0025] A-2. Configuration of information processing device 100: Next, a description will be given of the configuration of an information processing device 100 for executing the process of generating the driving risk profile RP. Fig. 2 is an explanatory diagram showing a schematic configuration of the information processing device 100. The information processing device 100 is configured by, for example, a computer (PC, server, etc.).

[0026] The information processing device 100 includes a control unit 110, a storage unit 120, a display unit 130, an operation input unit 140, and an interface unit 150. These units are connected to each other so as to be able to communicate with each other via a bus 190. The information processing device 100 may also include a speaker as an output means.

[0027] The display unit 130 of the information processing device 100 is configured, for example, by a liquid crystal display or the like, and displays various images and information. The operation input unit 140 is configured, for example, by a keyboard, mouse, buttons, a microphone, a trackpad, or the like, and receives operations and instructions from an administrator. The display unit 130 may also function as the operation input unit 140 by being equipped with a touch panel. The interface unit 150 is configured, for example, by a LAN interface, a USB interface, or the like, and communicates with other devices via wired or wireless connections.

[0028] The storage unit 120 of the information processing device 100 is configured with, for example, a ROM, a RAM, a hard disk drive (HDD), etc., and is used to store various programs and data, and as a work area when executing various programs, and as a temporary storage area for data. For example, the storage unit 120 stores a driver evaluation program CP, which is a computer program for executing a driver evaluation process described below. The driver evaluation program CP is provided in a state stored on a computer-readable recording medium (not shown), such as a CD-ROM, a DVD-ROM, or a USB memory, or is provided in a state that can be obtained from an external device (a server or other terminal device on a network) via the interface unit 150, and is stored in the storage unit 120 in a state that is operable on the information processing device 100.

[0029] Furthermore, acceleration data AD is stored in advance or during execution of a driver evaluation process, which will be described later, in the storage unit 120 of the information processing device 100. The contents of this data will be described together with the description of the driver evaluation process, which will be described later.

[0030] The control unit 110 of the information processing device 100 is configured with, for example, a CPU, and controls the operation of the information processing device 100 by executing a computer program read from the storage unit 120. For example, the control unit 110 functions as a driver evaluation processing unit 111 for executing a driver evaluation process described below by reading and executing a driver evaluation program CP from the storage unit 120. The driver evaluation processing unit 111 includes a data acquisition unit 112, a safe driving area setting unit 113, a classification processing unit 114, and a spatiotemporal information processing unit 117. The classification processing unit 114 includes a first clustering processing unit 115 and a second clustering processing unit 116. The functions of these units will be described in conjunction with the description of the driver evaluation process described below.

[0031] A-3. Driver evaluation process: Next, a driver evaluation process executed by the information processing device 100 of this embodiment will be described. Fig. 3 is a flowchart showing the driver evaluation process. The driver evaluation process is a process for generating the above-mentioned driving risk profile RP by evaluating the driving risk of the driver. The driver evaluation process is started in response to, for example, a user operating the operation input unit 140 of the information processing device 100 to input a start instruction.

[0032] First, the data acquisition unit 112 (FIG. 2) of the information processing device 100 acquires acceleration data AD for each of multiple drivers (S110). The acceleration data AD is data indicating a combination of lateral acceleration and vertical acceleration of a moving body (e.g., an automobile) at each timing when the driver is driving the moving body. The lateral acceleration is the acceleration in the left-right direction of the moving body and is correlated, for example, with the steering operation by the driver. The vertical acceleration is the acceleration in the front-rear direction of the moving body and is correlated, for example, with the accelerator and brake operation by the driver. The lateral acceleration and vertical acceleration of the moving body are measured by an acceleration sensor installed in the moving body, and the acceleration data AD indicating the measured values of the lateral acceleration and vertical acceleration are stored in a storage device (e.g., a drive recorder). In this embodiment, the acceleration data AD includes position information indicating the position of the moving body at each timing. The position information is measured and acquired by a position sensor, such as a GPS, installed in the moving body. The data acquisition unit 112 of the information processing device 100 acquires the acceleration data AD via the interface unit 150 at a predetermined timing.

[0033] Next, the data acquisition unit 112 of the information processing device 100 creates a GG diagram 20 for each of the multiple drivers based on the acceleration data AD (S120). FIG. 4 is an explanatory diagram showing an example of the GG diagram 20. As shown in FIG. 4, the GG diagram 20 is a graph with lateral acceleration on the horizontal axis and vertical acceleration on the vertical axis, showing the distribution of combinations of lateral acceleration and vertical acceleration of the mobile body at each timing when the driver is driving the mobile body. Note that creation of the GG diagram 20 based on the acceleration data AD may be performed on the mobile body side. In that case, the data acquisition unit 112 acquires the GG diagram 20 as acceleration data AD from the mobile body.

[0034] Next, the safe driving area setting unit 113 (FIG. 2) of the information processing device 100 sets a safe driving area SA on the GG diagram 20 (S130). FIG. 5 is an explanatory diagram showing an example of the safe driving area SA. The safe driving area SA is an area on the GG diagram 20 where a combination (data point) of lateral acceleration and vertical acceleration of a moving object that is recognized as a safe driving state is located. The safe driving area SA is set to distinguish between dangerous driving operations and safe driving operations. The safe driving area setting unit 113 sets the safe driving area SA using a statistical method based on the assumption that driving operations with a relatively low probability of occurrence are dangerous.

[0035] The safe driving area SA is set, for example, as follows: First, the plane of the GG diagram 20 is divided into a plurality of grid cells of a fixed size, and the density d of the data points in each grid cell is calculated. j is calculated according to the following formula (1). In formula (1), N j is the number of data points in grid cell j, and C is the number of grid cells in the plane of the GG diagram 20.

number

[0036] In this embodiment, one safe driving area SA is set common to all drivers. However, a safe driving area SA may be set for each driver attribute (for example, age group, vehicle type, region, etc.).

[0037] Next, the density of each grid cell, d j The cumulative sum D of the matrix d' is expressed as the following formula (3). In formula (3), D' j is the sum of the first j densities in matrix d'.

number

number

[0038] The q confidence region is the grid cell corresponding to the first k densities in the matrix d', where k is calculated according to the following equation (4). The q confidence region of the distribution of all drivers' data points is set as the safe driving area SA. In this embodiment, the 95% confidence region of the distribution of all drivers' data points is set as the safe driving area SA. This 95% value is set according to the 2σ rule and corresponds to μ±2σ.

number

[0039] As shown in Figure 4, the set safe driving area SA is a roughly diamond-shaped area. The reason the safe driving area SA has such a shape is that it is set based on a combination of lateral acceleration and longitudinal acceleration, which are correlated with each other. Compared to the conventional method of setting thresholds for each of the lateral acceleration and longitudinal acceleration and using a rectangular safe driving area specified by those thresholds, it is possible to set a safe driving area SA that more appropriately distinguishes between dangerous and safe driving operations.

[0040] Next, the first clustering processing unit 115 (FIG. 2) of the information processing device 100 executes the first clustering process C1 to classify the multiple drivers into multiple driving behavior groups DG with different driving risk levels (S140). This classification provides a more macroscopic representation of the heterogeneity among drivers. In this embodiment, as shown in FIG. 1, three driving behavior groups DG (DG1: high-risk driver group, DG2: medium-risk driver group, and DG3: low-risk driver group) are set, and each driver is classified into one of the three driving behavior groups DG. The number of driving behavior groups DG may be two, four, or more.

[0041] The first clustering process C1 is a process of classifying and labeling the average driving behavior of each driver into several groups. The first clustering process C1 is executed, for example, as follows. First, for each driver, two parameters are calculated based on the GG diagram 20. Specifically, a proportion Pr of risky driving behavior and an unstable driving index Ii are calculated. The proportion Pr of risky driving behavior is an example of a first parameter, and the unstable driving index Ii is an example of a second parameter.

[0042] The driving instability index Ii is a parameter that indicates the stability of acceleration operation, and the larger the value, the more unstable the driving. The driving instability index Ii is the average distance between all data points and their centers of gravity, and is calculated according to the following formula (5). In formula (5), x is the lateral acceleration, y is the vertical acceleration, and (x i ,y i ) represents data point i in the GG diagram 20, and (x c ,y c ) represents the centroid of all data points.

number

[0043] The proportion Pr of risky driving behavior is a parameter that indicates the frequency of risky driving behaviors, and is calculated, for example, as the proportion of the number of data points that are outside the safe driving area SA to the total number of data points.

[0044] The first clustering processing unit 115 executes a first clustering process C1 to classify multiple drivers using two parameters: a driving instability index Ii and a risky driving behavior rate Pr. The first clustering process C1 uses, for example, a hierarchical clustering algorithm, which is an unsupervised machine learning algorithm. In this embodiment, the algorithm employs an agglomerative approach. In this agglomerative approach, each of multiple data points is initially treated as a separate cluster, and two clusters are merged into one cluster based on the pairwise similarity between the two clusters. This process is repeated until all data points are merged into one cluster. The final number of clusters can be determined using various methods, including, for example, a silhouette index indicating how close each data point is to its own cluster compared to other clusters. Figure 6 is an explanatory diagram illustrating an example of the results of the first clustering process C1. As shown in Figure 6, in this embodiment, the first clustering process C1 classifies each driver into one of the three driving behavior groups DG described above.

[0045] Next, the second clustering processing unit 116 (FIG. 2) of the information processing device 100 executes the second clustering process C2 to identify each driver's risky driving behavior pattern DP (S150). Specifically, the second clustering processing unit 116 breaks down each driver's risky driving behavior into several risk factors, classifies each factor into multiple risky driving behavior patterns DP, and identifies the proportion of each of the multiple risky driving behavior patterns DP. This classification is intended to diagnose each driver's individual driving behavior and provides more detailed information about the heterogeneity of each driver's driving style. The classification in the second clustering process C2 and the classification in the first clustering process C1 described above are complementary, and the two classifications allow a more comprehensive understanding of each driver's driving style.

[0046] In this embodiment, a Gaussian Mixture Model (abbreviated as GMM) is used in the second clustering process C2. The Gaussian Mixture Model is a model expressed as a weighted sum of a finite number of probability density functions, as shown in the following formula (6). In formula (6), x is a two-dimensional feature vector, and p(x|μ i ,Σ i ) (where i=1,2,...,K) is the probability density function of the i-th element, and α i is the weight of the i-th element (where α1+α2+···+α K =1). Each element is the mean vector μ i and covariance matrix Σ i is a two-dimensional Gaussian distribution with

number

[0047] In the second clustering process C2, a Gaussian mixture model is trained individually for each driver. The parameters are estimated using, for example, the expectation maximization (EM) algorithm. The number of elements in the Gaussian mixture model is determined based on, for example, the Bayesian information criterion (BIC) score. The BIC score is calculated according to the following formula (7): In formula (7), L(ψ) is the observed likelihood of the mode, k is the number of parameters in the model, and N is the number of training data points. A lower BIC score indicates better performance of the estimated model. For example, the number of elements is increased until the BIC score of the model does not substantially decrease even with an increase in the number of elements.

number

[0048] Each element of the Gaussian mixture model becomes a cluster, and the Gaussian density μ i The mean vector of represents the center of gravity of the cluster. i represents the percentage of data points clustered in this group. Therefore, the weight α i can be interpreted as the probability of one random risky operation belonging to this cluster.

[0049] The center of gravity of each element is used to label each element. In this embodiment, a statistical method is used to determine two thresholds according to the 10th percentile of the magnitude of lateral and longitudinal acceleration, and eight risky driving behavior patterns DP are defined. Table 1 shows the definitions of the eight risky driving behavior patterns DP. In this embodiment, the above-mentioned eight risky driving behavior patterns DP are set according to the following definitions. [Table 1]

[0050] Table 2 shows the Gaussian mixture model estimated for a driver. In the example in Table 2, the Gaussian mixture model consists of 10 elements, and the weight of each element and the center of gravity for labeling are specified. [Table 2]

[0051] For each driver, the proportion of risky driving behaviors belonging to each of multiple risky driving behavior patterns DP (see Figure 1) is identified based on the weight of each element of the Gaussian mixture distribution model and the label of each element (risky driving behavior pattern DP).

[0052] In this embodiment, the second clustering processor 116 further calculates a normalized driving performance score PS for each driver based on the Gaussian mixture distribution model estimated for each driver. First, the risk index R of the driver m is calculated according to the following equation (8): m In equation (8), P m is the proportion of risky driving maneuvers by driver m, and |μ i | is the centroid μ of the i-th element i =(a xc ,a yc ) and is calculated according to the following formula (9):

number

number

[0053] Normalized driving performance score PS from 0 to 100 m is calculated according to the following formula (10). In formula (10), R baseline is the risk index R of the driver belonging to the safest driving behavior group DG (i.e., DG3). mIf the calculated value according to formula (10) exceeds 100, PS is set to 100. Therefore, a driving performance score PS of 100 does not indicate no risk, but rather indicates a driving style that is safer than the average level of the safest driving behavior group DG.

number

[0054] Next, the spatiotemporal information processing unit 117 (FIG. 2) of the information processing device 100 creates spatiotemporal information STI of risky driving behaviors for each driver (S160). The spatiotemporal information STI is information indicating the time and location of the risky driving behaviors by the driver. As described above, in this embodiment, the acceleration data AD includes location information indicating the location of the mobile body at each timing while the driver is driving the mobile body. Therefore, by referring to the acceleration data AD, the location of the risky driving behaviors by the driver can be identified. The location of the risky driving behaviors may be identified as a point based on a combination of latitude and longitude, for example, or may be identified by type of location (e.g., intersection or segment). FIGS. 7 and 8 are explanatory diagrams showing an example of how the spatiotemporal information is used. FIG. 7 shows the frequency of risky driving behaviors by a certain driver by time and by location (intersection or segment). FIG. 8 also shows the location HP of the risky driving behavior superimposed on a road map. By referencing spatiotemporal information STI, we can identify the times and places (hotspots) where risky driving behavior frequently occurs for each driver. This will enable us to clarify the implicit factors behind risky driving behavior and enable a fairer comparison of driver performance. Furthermore, by referencing spatiotemporal information STI, we can improve the interpretability of models and evaluation results.

[0055] The driver evaluation processing unit 111 (FIG. 2) of the information processing device 100 compiles the results of the above-mentioned processes to create a driving risk profile RP (FIG. 1) for each driver, and displays the driving risk profile RP on the display unit 130 (S170). FIG. 9 is an explanatory diagram showing another example of a driving risk profile RP. FIG. 9 shows an individual driving risk profile RP for a certain driver. The driving risk profile RP shown in FIG. 9 includes a driver ID that identifies the driver, a driving behavior group DG to which the driver belongs, a driving performance score PS calculated for the driver, and a pie chart showing the proportion of risky driving behaviors belonging to each of a plurality of risky driving behavior patterns DP. By referring to the driving risk profile RP, it is possible to understand the driver's driving risk characteristics in more detail.

[0056] A-4. Advantages of this embodiment: As described above, the information processing device 100 of this embodiment includes a data acquisition unit 112 and a classification processing unit 114. The data acquisition unit 112 acquires, for a plurality of drivers, acceleration data AD indicating combinations of lateral acceleration and vertical acceleration of the mobile body at each timing when the mobile body is being driven. The classification processing unit 114 executes a second clustering process C2 using the acceleration data AD for each of the plurality of drivers, thereby classifying a plurality of risky driving behaviors by the drivers into a plurality of risky driving behavior patterns DP set based on the combinations of lateral acceleration and vertical acceleration, and outputs the classification results.

[0057] In this way, the information processing device 100 of this embodiment outputs the results of classifying multiple risky driving behaviors by a driver into multiple risky driving behavior patterns DP set based on combinations of lateral acceleration and longitudinal acceleration, thereby making it possible to identify the driver's driving risk characteristics in more detail. Furthermore, the information processing device 100 of this embodiment can evaluate driving risk characteristics using large-scale unlabeled driving history data (e.g., data recorded by a drive recorder), making it easy to collect data and reducing the workload of manually labeling data and having evaluators evaluate drivers individually.

[0058] In this embodiment, the classification result by the information processing device 100 includes information indicating the proportion of risky driving behaviors belonging to each risky driving behavior pattern DP. Therefore, more detailed driving risk characteristics of the driver can be identified in a more easily understandable manner.

[0059] In this embodiment, the classification processing unit 114 executes the second clustering process C2 using a Gaussian mixture distribution model, which allows for more detailed driving risk characteristics of the driver to be identified with high accuracy.

[0060] In this embodiment, the acceleration data AD includes location information indicating the location of the mobile body at each timing while the mobile body is being driven, and the information processing device 100 further includes a time-space information processing unit 117 that generates time-space information STI indicating the occurrence time and location of risky driving behavior and outputs the time-space information STI. Therefore, for each driver, it is possible to identify the times and places (hot spots) where risky driving behavior frequently occurs, which in turn makes it possible to clarify the implicit factors behind the driver's risky driving behavior and to more fairly compare the driving performance of drivers.

[0061] The information processing device 100 of this embodiment also includes a data acquisition unit 112 and a classification processing unit 114. The data acquisition unit 112 acquires acceleration data AD indicating a combination of lateral acceleration and vertical acceleration of a mobile body at each timing when the mobile body is being driven for a plurality of drivers. The classification processing unit 114 executes a first clustering process C1 using the acceleration data AD to classify the plurality of drivers into a plurality of driving behavior groups DG having different driving risk levels, and outputs the classification results.

[0062] In this way, the information processing device 100 of this embodiment outputs the results of classifying multiple drivers into multiple driving behavior groups DG with different driving risk levels, and therefore can identify more detailed driving risk characteristics of the driver.

[0063] In this embodiment, the classification processing unit 114 calculates a first parameter (proportion Pr of risky driving behavior) indicating the proportion of driving operations corresponding to predetermined dangerous acceleration operations and a second parameter (driving instability index Ii) indicating the instability of driving behavior based on the acceleration data AD, and executes a first clustering process C1 using the first parameter and the second parameter. Therefore, more detailed driving risk characteristics of the driver can be identified with high accuracy.

[0064] In this embodiment, the classification processing unit 114 executes the first clustering process C1 using a hierarchical clustering algorithm, which allows for more detailed driving risk characteristics of the driver to be identified with high accuracy.

[0065] B. Variations: The technology disclosed in this specification is not limited to the above-described embodiments, and can be modified into various forms without departing from the spirit thereof, for example, the following modifications are also possible.

[0066] The configuration of the information processing device 100 in the above embodiment is merely an example and can be modified in various ways. Furthermore, the contents of the driver evaluation process in the above embodiment are merely an example and can be modified in various ways. For example, in the above embodiment, the information processing device 100 executes the first clustering process C1 and the second clustering process C2, but the information processing device 100 may execute only one of the first clustering process C1 and the second clustering process C2.

[0067] The data contents and processing algorithms used in the driver evaluation process in the above embodiment are merely examples and can be modified in various ways.

[0068] In the above-described embodiment, a part of the configuration realized by hardware may be replaced by software, and conversely, a part of the configuration realized by software may be replaced by hardware. [Explanation of symbols]

[0069] 20: GG diagram 100: Information processing device 110: Control unit 111: Driver evaluation processing unit 112: Data acquisition unit 113: Safe driving area setting unit 114: Classification processing unit 115: First clustering processing unit 116: Second clustering processing unit 117: Spatiotemporal information processing unit 120: Memory unit 130: Display unit 140: Operation input unit 150: Interface unit 190: Bus AD: Acceleration data CP: Driver evaluation program

Claims

1. An information processing device, a data acquisition unit that acquires acceleration data indicating a combination of lateral acceleration and vertical acceleration of the moving body at each timing when the moving body is being driven by a plurality of drivers; a classification processing unit that performs clustering using the acceleration data for each of the plurality of drivers to classify a plurality of risky driving behaviors of the drivers into a plurality of risky driving behavior patterns that are set based on combinations of the lateral acceleration and the longitudinal acceleration, and outputs the classification results; An information processing device comprising:

2. 2. The information processing device according to claim 1, An information processing device, wherein the classification result includes information indicating the proportion of the risky driving behaviors that belong to each of the risky driving behavior patterns.

3. 3. The information processing device according to claim 1, The information processing device, wherein the classification processing unit performs the clustering using a Gaussian mixture distribution model.

4. 3. The information processing device according to claim 1, the acceleration data includes position information indicating a position of the moving object at each of the timings, The information processing device further includes a spatiotemporal information processing unit that creates spatiotemporal information indicating the occurrence time and occurrence position of the risky driving behavior and outputs the spatiotemporal information.

5. An information processing device, a data acquisition unit that acquires acceleration data indicating a combination of lateral acceleration and vertical acceleration of the moving body at each timing when the moving body is being driven by a plurality of drivers; a classification processing unit that classifies the plurality of drivers into a plurality of driving behavior groups having different driving risk levels by performing clustering using the acceleration data, and outputs the classification results; An information processing device comprising:

6. 6. The information processing device according to claim 5, The classification processing unit calculates, based on the acceleration data, a first parameter indicating the proportion of driving operations that correspond to predetermined dangerous acceleration operations and a second parameter indicating the instability of driving behavior, and performs the clustering using the first parameter and the second parameter.

7. 7. The information processing device according to claim 5, The classification processing unit performs the clustering using a hierarchical clustering algorithm.

8. 7. The information processing device according to claim 5, The classification processing unit further performs clustering using the acceleration data for each of the plurality of drivers, thereby classifying the plurality of risky driving behaviors of the drivers into a plurality of risky driving behavior patterns set based on combinations of the lateral acceleration and the longitudinal acceleration, and outputs the classification results.

9. An information processing method, comprising: acquiring acceleration data indicating a combination of lateral acceleration and vertical acceleration of the moving body at each timing when the moving body is being driven by a plurality of drivers; a step of classifying a plurality of risky driving behaviors of the drivers into a plurality of risky driving behavior patterns set based on combinations of the lateral acceleration and the longitudinal acceleration by performing clustering using the acceleration data for each of the plurality of drivers, and outputting the classification results; An information processing method comprising:

10. An information processing method, comprising: acquiring acceleration data indicating a combination of lateral acceleration and vertical acceleration of the moving body at each timing when the moving body is being driven by a plurality of drivers; classifying the plurality of drivers into a plurality of driving behavior groups having different driving risk levels by performing clustering using the acceleration data, and outputting the classification results; An information processing method comprising:

11. A computer program comprising: On the computer, A process of acquiring acceleration data indicating a combination of lateral acceleration and vertical acceleration of a moving body at each timing when the moving body is being driven by a plurality of drivers; a process of classifying a plurality of risky driving behaviors of the drivers into a plurality of risky driving behavior patterns set based on combinations of the lateral acceleration and the longitudinal acceleration by performing clustering using the acceleration data for each of the plurality of drivers, and outputting the classification results; A computer program that executes

12. A computer program comprising: On the computer, A process of acquiring acceleration data indicating a combination of lateral acceleration and vertical acceleration of a moving body at each timing when the moving body is being driven by a plurality of drivers; a process of classifying the plurality of drivers into a plurality of driving behavior groups having different driving risk levels by performing clustering using the acceleration data, and outputting the classification results; A computer program that executes

Images