JP2019153257A - Abnormal driving action detection device, abnormal driving action detection method and program - Google Patents

Abstract

To provide an abnormal driving action detection device, abnormal driving action detection method and program, capable of detecting an abnormal driving action concerning acceleration and deceleration operations by a driver with a simple configuration.SOLUTION: An abnormal driving action detection device 1 includes: vehicle acquisition means 12 for acquiring the vehicle speed of a vehicle 2; a storage section 11; vehicle speed estimation means 13; and abnormal action detection means 14. The storage section 11 stores information showing a vehicle speed acquired by the vehicle speed acquisition means 12 and a predetermined autoregression model M. The vehicle speed estimation means 13 estimates a vehicle speed at a predetermined point of time on the basis of the autoregression model M and a plurality of vehicle speeds at points of time different from each other, earlier than the predetermined point of time. The abnormal action detection means 14 detects an abnormal action when a difference between the vehicle speed acquired by the vehicle speed acquisition means 12 and the vehicle speed at the predetermined point of time estimated by the vehicle speed estimation means 13 is in excess of a predetermined threshold value.SELECTED DRAWING: Figure 1

Description

  The present invention relates to an abnormal driving behavior detection device, an abnormal driving behavior detection method, and a program for detecting abnormal behavior related to acceleration / deceleration operations by a driver driving a vehicle.

  As a safety measure in driving a vehicle, a drive recorder (Patent Documents 1 and 2) that detects an abnormality with an acceleration sensor or a magnetic sensor and images the periphery of the vehicle with a surveillance camera, a vehicle speed sensor, a yaw rate sensor, a brake pressure sensor, etc. A technique (Patent Document 3) that detects abnormalities by various sensors and performs safe driving support is known.

JP 2007-88541 A JP 2007-66194 A JP 2009-29343 A

  In the techniques disclosed in Patent Documents 1 to 3, it is necessary to mount a large number of sensors for detecting an abnormality in the vehicle, so that the abnormality detecting system is complicated.

  The present invention has been made in view of the above circumstances, and has an abnormal driving behavior detection device, an abnormal driving behavior detection method, and a program capable of detecting abnormal behavior related to acceleration / deceleration operations by a driver with a simple configuration. The purpose is to provide.

In order to achieve the above object, an abnormal driving behavior detection apparatus according to the first aspect of the present invention includes:
An abnormal driving behavior detection device that detects abnormal behavior related to acceleration / deceleration operations by a driver driving a vehicle,
Obtaining means for obtaining a vehicle speed of the vehicle;
Storage means for storing information indicating the vehicle speed acquired by the acquisition means and a predetermined autoregressive model for estimating the vehicle speed at a predetermined time;
Estimating means for estimating the vehicle speed at the predetermined time point based on the autoregressive model and a plurality of vehicle speeds at different time points before the predetermined time point;
And detecting means for detecting the abnormal behavior when a difference between the vehicle speed acquired by the acquiring means and the vehicle speed at the predetermined time estimated by the estimating means exceeds a predetermined threshold.

  The detecting means detects the abnormal behavior when the difference between the vehicle speed acquired by the acquiring means at a time before the predetermined time and the vehicle speed at the predetermined time estimated by the estimating means exceeds the threshold. You may make it detect.

The order of the autoregressive model is predetermined,
The estimation means may estimate the vehicle speed at the predetermined time point by substituting the vehicle speed for the order at different time points before the predetermined time point into the autoregressive model.

  The autoregressive model may be determined in advance based on a time series of vehicle speeds when the vehicle is driven a plurality of times on a predetermined route in a state where the abnormal behavior does not occur.

  The vehicle may be a virtual vehicle by a simulator, and the abnormal driving behavior detection device may be provided in the simulator.

In order to achieve the above object, an abnormal driving behavior detection method according to the second aspect of the present invention includes:
An abnormal driving behavior detection method for detecting abnormal behavior related to acceleration / deceleration operations by a driver driving a vehicle,
An acquisition step of acquiring a vehicle speed of the vehicle;
An estimation step for estimating the vehicle speed at the predetermined time point based on a predetermined autoregressive model for estimating the vehicle speed at the predetermined time point and a plurality of vehicle speeds at different time points before the predetermined time point;
A detection step of detecting the abnormal behavior when a difference between the vehicle speed acquired in the acquisition step and the vehicle speed at the predetermined time point estimated in the estimation step exceeds a predetermined threshold value.

In order to achieve the above object, a program according to the third aspect of the present invention provides:
A program for detecting abnormal behavior related to acceleration / deceleration operations by a driver driving a vehicle,
Computer
Obtaining means for obtaining the vehicle speed of the vehicle;
Estimating means for estimating the vehicle speed at the predetermined time point based on a predetermined autoregressive model for estimating the vehicle speed at the predetermined time point and a plurality of vehicle speeds at different time points before the predetermined time point;
Detecting means for detecting the abnormal behavior when the difference between the vehicle speed acquired by the acquiring means and the vehicle speed at the predetermined time estimated by the estimating means exceeds a predetermined threshold;
To function as.

  According to the present invention, it is possible to detect an abnormal behavior related to an acceleration / deceleration operation by a driver with a simple configuration.

It is a block diagram which shows the structure of the abnormal driving action detection apparatus which concerns on one Embodiment of this invention. It is a figure which shows the structure of the autoregressive model which concerns on one Embodiment of this invention. It is a figure which shows an example of the course used for driving action measurement. It is a figure of the graph which shows an example of the vehicle speed actual value in the course shown in FIG. It is a figure for demonstrating the estimation precision of the autoregressive model which concerns on one Embodiment of this invention. It is a figure of the graph which shows the difference of the actual value of a vehicle speed, and an estimated value. It is a flowchart which shows the abnormal action detection process which concerns on one Embodiment of this invention.

  An embodiment of the present invention will be described with reference to the drawings.

  As shown in FIG. 1, the abnormal driving behavior detection device 1 according to the present embodiment is mounted on a vehicle 2 that is an automobile, for example, or is provided in a simulator that enables driving of a virtual vehicle 2. It is done.

  The abnormal driving behavior detection device 1 is applicable regardless of whether the vehicle 2 is a real vehicle or a virtual vehicle. Therefore, in the following, the vehicle 2 is a real vehicle unless otherwise noted. Various configurations will be described without distinguishing between vehicles and virtual vehicles.

  The abnormal driving behavior detection device 1 detects abnormal behavior related to an acceleration / deceleration operation by a driver driving the vehicle 2, and includes a control unit 10 and a notification unit 20.

  The control unit 10 includes a microcomputer and controls the operation of each unit of the abnormal driving behavior detection device 1. The control unit 10 includes a storage unit 11 including a ROM (Read Only Memory) that stores operation programs and fixed data, a RAM (Random Access Memory) that temporarily stores various calculation results, and the like, and a CPU (Central Processing). Unit), a drive circuit for driving the notification unit 20, an input / output interface, and the like. In the ROM of the storage unit 11, as will be described later, data of various operation programs including a program P for executing abnormal behavior detection processing, data indicating an autoregressive model M, and the like are stored in advance. Hereinafter, the autoregressive model is also referred to as an “AR (AutoRegressive) model”.

  The control unit 10 includes functions as a vehicle speed acquisition unit 12, a vehicle speed estimation unit 13, and an abnormal behavior detection unit 14.

  When the vehicle 2 is an actual vehicle, the vehicle speed acquisition unit 12 acquires a vehicle speed signal S from the vehicle speed sensor 3 via, for example, a CAN (Controller Area Network), and calculates the vehicle speed based on the acquired vehicle speed signal S. The vehicle speed sensor 3 mounted on the vehicle 2 includes, for example, a hall element that detects a detected portion that rotates in synchronization with a wheel (for example, gear irregularities or metal protrusions), and a vehicle speed signal S having a frequency corresponding to the vehicle speed. Is supplied to the control unit 10. The control unit 10 functions as the vehicle speed acquisition unit 12 and performs A / D (Analog to Digital) conversion on the acquired vehicle speed signal S, and calculates (acquires) the vehicle speed according to the frequency of the vehicle speed signal S in a predetermined control cycle. And stored in the storage unit 11. In the case where the vehicle 2 is a virtual vehicle by a simulator, the vehicle speed acquisition unit 12 may acquire the vehicle speed according to the amount of operation of the accelerator pedal by the user (subject) of the simulator, for example.

  The vehicle speed estimation means 13 estimates the vehicle speed at a predetermined time based on the vehicle speed acquired by the vehicle speed acquisition means 12 and the AR model M stored in advance in the storage unit 11.

Here, the time series AR model M having the structure shown in FIG. 2 in the present embodiment will be described. Time series information indicating the vehicle speed at a certain time t is V (t). For time series information before (past) before V (t), the time series information before one sample is V (t) · z ⁻¹ and is multiplied by a coefficient a ₁ . Similarly, the time-series information before n samples and V (t) · ^{z -n,} multiplied by a coefficient _{a n.} The AR model M is configured by summing these past time series information to V (t) as an output Var (t) at the next sample time of V (t). The AR model M is expressed by the following formula.
Var (t) = V (t) + a ₁ · V (t) · z ⁻¹ +... + _An • V (t) · z ⁻ⁿ
The model order (maximum value of n) of the AR model M is determined based on the Akaike's Information Criterion (AIC: Akaike's Information Criterion), and a value at which the model output value is saturated when the order is increased is determined and applied.

Determining model order n and coefficients _a 1 ~a _n of AR model M is the vehicle speed obtained by a course as shown in FIG. 3 (path from Start to Goal), thereby driving the vehicle 2 to a plurality of driver This is based on the actual measurement value.

  As an example, the inventors of the present application constructed the course shown in FIG. 3 on a driving simulator and measured driving behavior by driving a virtual vehicle 2 by a plurality of subjects, that is, actually measured vehicle speed. In addition, although the mark of “x” shown in FIG. 3 indicates the occurrence location of “jumping stimulus” in which unexpected jumping of a person or a vehicle is set, when the AR model M is constructed, the jumping stimulus is given to the course. The subject is caused to drive the vehicle 2 in a state where it is not generated at all (that is, in a normal state where the probability of occurrence of abnormal driving behavior is low).

  In FIG. 4, the graph of the time series information of the vehicle speed actual value by a test subject is shown. The graph is obtained by plotting V (t) as a vehicle speed actual measurement value for each sample time, and the sample time is 0.01 [s]. Based on the AIC, the model order and coefficient of the AR model M are determined in advance from the time-series information of the vehicle speed actual measurement values for the number of subjects obtained in this way. In the example by simulation, the model order n is suitable for the 7th order based on the measured values of more than a dozen subjects (n = 7).

The vehicle speed estimation means 13 includes the AR model M determined as described above, V (t) as vehicle speed information currently acquired by the vehicle speed acquisition means 12 (at time t), and the time stored in the storage unit 11. Substitute the vehicle speed (V (t) · z ⁻¹ ,..., V (t) · z ⁻ⁿ ) for n samples before t, and estimate the vehicle speed Var at the next sample time after time t. (T) is calculated. That is, the vehicle speed at the next sample time is estimated.

  Here, FIG. 5 shows an estimation of the vehicle speed calculated by the vehicle speed estimation means 13 based on the measured value of the vehicle speed by simulation in the course shown in FIG. 3 and the AR model M in which the model order n is 7th order (n = 7). The graph of the comparison result with a value is shown. In the graph shown in FIG. 5, the behaviors of the two coincide with each other so that the measured value and the estimated value of the vehicle speed cannot be distinguished. Thereby, it can be seen that the AR model M has sufficient estimation accuracy. Note that seven samples corresponding to n = 7 correspond to 0.07 [s]. In this description, the course shown in FIG. 3 is constructed on the driving simulator, the vehicle speed is measured by causing a plurality of subjects to drive the virtual vehicle 2, and the AR model M is determined based on the measured vehicle speed. However, the AR model M is based on the time series of vehicle speeds when a predetermined actual course (route) is run a plurality of times in a state where abnormal behavior related to the acceleration / deceleration operation by the driver is not generated. Of course, it may be determined.

  The abnormal behavior detection means 14 includes the vehicle speed actual measurement value V (t) acquired by the vehicle speed acquisition means 12 at time t, and the vehicle speed estimation value Var (t) estimated by the vehicle speed estimation means 13 based on the AR model M as described above. Is detected in advance, the abnormal behavior related to the acceleration / deceleration operation by the driver is detected.

  Here, a specific example of the abnormality detection method when the course shown in FIG. 3 is constructed on the simulator will be described. As shown in FIG. 3, the vehicle speed when the subject drives the course is measured while generating a jumping stimulus at a plurality of locations marked with “X” on the course. The vehicle speed actual measurement value is a vehicle speed acquired by the vehicle speed acquisition unit 12 at every predetermined sample time, and is stored in the storage unit 11. Note that whether or not a jumping out stimulus occurs at each of the plurality of locations is randomly set for each subject. That is, when the actual vehicle 2 is driven on a predetermined route, the vehicle speed is actually measured in consideration of popping out of a person or vehicle that may occur at random. Therefore, the abnormality detection method can be applied regardless of whether the vehicle 2 is a real vehicle or a virtual vehicle.

  Further, the vehicle speed estimating means 13 obtains an actual measured value V (t) of the vehicle speed based on the vehicle speed acquired by the vehicle speed acquiring means 12 at every predetermined sample time and stored in the storage unit 11 and a predetermined AR model M. A vehicle speed estimated value Var (t) at the next sample time after the acquired time t is calculated.

  As described above, since the AR model M is constructed without causing any jump-out stimulus on the course, the time series of the vehicle speed estimation value Var (t) calculated based on the AR model M is abnormal driving. It can be assumed that the normal state where the probability that the action occurs is low. Note that the instantaneous driving behavior of sudden acceleration or sudden deceleration included in the actual measurement value V (t) for calculating the vehicle speed estimated value Var (t) is very small relative to the entire driving behavior. The time series of the vehicle speed estimated value Var (t) calculated based on the model M can be regarded as a normal state.

  On the other hand, if the actual measurement value V (t) includes the result of sudden acceleration or deceleration as little as possible, when the difference between the actual measurement value V (t) and the estimated vehicle speed Var (t) is obtained, instantaneous driving is performed. The difference is maximized in the vicinity of the timing of the action. The abnormal behavior detecting means 14 detects the abnormal behavior deviating from the normal state by detecting the occurrence of the maximum value with a predetermined threshold.

  FIG. 6 shows a time-series graph of the difference between the measured value V (t) of the vehicle speed and the estimated value Var (t). T1 in the figure is an example of a threshold value for detecting a maximum value when the difference is considered as an absolute value, and T2 is an example of an abnormal threshold value for detecting abnormal behavior. When T1 and T2 are considered as absolute values, T2> T1. In the figure, the location where the difference exceeds the abnormal threshold T2 is a location where a jumping stimulus is given or a location where acceleration / deceleration occurs due to a signal change on the course, and abnormal behavior by the driver is detected. I understand that. Note that the threshold value T1 and the abnormal threshold value T2 for detecting the local maximum value shown in the figure are merely examples, and appropriate values can be arbitrarily determined through experiments.

  When the abnormal behavior detecting means 14 detects the abnormal behavior, the notification unit 20 notifies the driver of the vehicle 2 with an image, sound, or the like that the acceleration / deceleration operation is abnormal. As the notification unit 20, an image display device composed of an LCD (Liquid Crystal Display), an OLED (Organic Light Emitting Diodes), a speaker, or the like can be employed. For example, acceleration / deceleration operations such as “accelerator operation, danger!” Are performed. Notification is performed by displaying an image indicating an abnormality, or outputting a sound or warning sound indicating the same content.

  The description of the configuration of the abnormal driving behavior detection device 1 has been described above. Subsequently, the abnormal action detection process executed by the control unit 10 according to the program P stored in the storage unit 11 by the control unit 10 will be described with reference to FIG.

  When the abnormal behavior detection process is started, the vehicle speed acquisition unit 12 of the control unit 10 acquires the actual measured value V (t) of the vehicle speed at the current time t and stores it in the storage unit 11 (step S1). By repeatedly executing the abnormal behavior detection process, the actual measurement value V (t) at each predetermined sample time is stored in the storage unit 11.

Subsequently, the vehicle speed estimation means 13 of the control unit 10 determines the actual measurement value V (t) acquired at time t and the past actual measurement value V (t) · z ⁻¹ ,. Substituting V (t) · z ⁻ⁿ into the AR model M, the estimated vehicle speed value Var (t) at the next sample time after time t is calculated and stored in the storage unit 11 (step S2). For example, when the order is 7 (n = 7), the actual measurement value V (t) from 1 sample before to 7 samples before time t is substituted into the AR model M. If the past measured value V (t) for the order n has not yet been stored in the storage unit 11, for example, step S1 is repeatedly executed or the process is skipped from step S1 to step S5 described later. You can do it.

  Subsequently, the abnormal behavior detecting unit 14 of the control unit 10 calculates a difference (absolute value) between the actual measurement value V (t) acquired in step S1 and the vehicle speed estimation value Var (t) calculated in step S2 (step). S3).

  Subsequently, the abnormal behavior detecting unit 14 determines whether or not the difference calculated in Step S3 indicates a maximum value based on the threshold value T1 stored in the storage unit 11 in advance (Step S4). Specifically, when the calculated difference is equal to or less than the threshold value T1, the abnormal behavior detecting unit 14 determines that the difference does not indicate the maximum value (Step S4; No), and that there is no abnormality in the driving behavior. Is detected (step S5).

  On the other hand, when the calculated difference is larger than the threshold value T1, the abnormal behavior detecting unit 14 determines that the difference indicates the maximum value (step S4; Yes), and proceeds to step S6.

  In step S6, the abnormal behavior detecting unit 14 determines whether or not the difference indicating the maximum value calculated in step S3 is larger than the abnormal threshold T2 (T2> T1) stored in the storage unit 11 in advance. When the difference is equal to or less than the abnormality threshold T2 (step S6; No), it is detected that there is no abnormality in the driving behavior (step S5). On the other hand, when the difference is larger than the abnormality threshold T2 (step S6; Yes), the abnormality of the driving behavior is detected (step S7), and the presence of the abnormality is notified to the driver by the notification unit 20. The control unit 10 repeatedly executes the above abnormal behavior detection processing until an end instruction is given.

  The present invention is not limited to the above embodiments and drawings. Changes (including deletion of components) can be made as appropriate without departing from the scope of the present invention. Examples of various modifications will be described below.

  When it is detected in step S5 of the abnormal behavior detection process that there is no abnormality in the driving behavior, the notification unit 20 may notify that there is no abnormality (that is, normal driving). Further, when the maximum value is larger than the abnormality threshold value T2 in step S6 (step S6; No), the warning state as the previous stage of abnormality is detected without determining that there is no abnormality in the driving behavior (step S5). You may do it. The notification unit 20 may display an image indicating that the warning is generated, for example, an image indicating content such as “Let's drive with caution”, or may output a sound or warning sound indicating the content.

  In the abnormal behavior detection process described above, the example in which the threshold value T1 for detecting the maximum value is provided has been described. However, the process of step S4 is deleted, and the determination process of step S6 is executed following step S3. It may be. That is, it may be determined whether or not the driving behavior is abnormal using only the abnormal threshold value T2 without using the threshold value T1.

  In the above abnormal behavior detection process, an example in which an abnormality is detected based on the difference between the actually measured value V (t) acquired at time t and the estimated value Var (t) of the vehicle speed at the next sample time after time t. However, the present invention is not limited to this. As described above, the time series of the estimated values calculated based on the autoregressive model M can be regarded as a normal state, and there is little rapid change, and the sampling period is as short as 0.01 [s], for example. Since it is an interval, it is calculated based on the actual measurement value V (t) acquired at time t and the estimated value Var (t) of the vehicle speed at time t (that is, based on a plurality of actual measurement values acquired at the sample time before time t. Abnormality may be detected based on the difference from the estimated value). In addition, the actual measurement value V (t) acquired at time t and the estimated vehicle speed Var (t) at the time several samples before time t (that is, the sample time before the time several samples earlier) An anomaly may be detected based on a difference from an estimated value calculated based on a plurality of actual measurement values acquired in (1). However, in order to bring the point of abnormality detection as close to real time as possible, the difference between the actually measured value V (t) acquired at time t and the estimated value Var (t) of the vehicle speed at the next sample time after time t is used. It is preferable to detect an abnormality based on this.

Further, the order of the autoregressive model M is 7th order suitable in the simulation example, but it is preferably about 5-7th order, and other orders are arbitrary depending on the purpose. is there. Further, the autoregressive model M is not limited to the example shown in the above formula, and a known time series analysis method can be adopted as appropriate, and how to construct it is arbitrary. For example, coefficients other than the coefficient to be multiplied on the past vehicle speed measured value (a 1 ^{~a n),} it may be such as to consider the white noise.

  Further, the vehicle 2 on which the abnormal driving behavior detection device 1 is mounted and the virtual vehicle 2 by the provided simulator are arbitrary, and are not limited to an automobile, but may be a motorcycle. The abnormal driving behavior detection device 1 can detect an abnormality with a simple system configuration based on the output of the vehicle speed sensor 3, but the vehicle 2 is also provided with other various sensors (camera, acceleration sensor, etc.) You may comprise the abnormality detection system which used these sensors together.

  Although the operation program (program P) for executing each process described above is stored in the storage unit 11 in advance, it may be distributed and provided by a removable recording medium. Further, the program P may be downloaded from another device connected to the abnormal driving behavior detection device 1. Moreover, the abnormal driving behavior detection apparatus 1 may execute each process according to the program P by exchanging various data with other devices via a telecommunication network or the like.

(1) The abnormal driving behavior detection device 1 described above detects abnormal behavior related to an acceleration / deceleration operation by a driver driving the vehicle 2, and a vehicle speed acquisition unit 12 (acquisition unit) and a storage unit 11 (storage unit). Vehicle speed estimation means 13 (estimation means) and abnormal behavior detection means 14 (detection means).
The vehicle speed acquisition unit 12 acquires the vehicle speed of the vehicle 2. The storage unit 11 stores information indicating the vehicle speed acquired by the vehicle speed acquisition unit 12 and a predetermined autoregressive model M for estimating the vehicle speed (Var (t)) at a predetermined time point. The vehicle speed estimation means 13 includes a plurality of vehicle speeds (V (t), V (t) · z ⁻¹ ,..., V (t) · z at different time points before the predetermined time point. ^-N ) to estimate the vehicle speed at a predetermined time point. The abnormal behavior detecting unit 14 is configured such that a difference between the vehicle speed (V (t)) acquired by the vehicle speed acquiring unit 12 and the vehicle speed (Var (t)) at a predetermined time estimated by the vehicle speed estimating unit 13 is a predetermined threshold T2. Abnormal behavior is detected when exceeding.

  As described above, since the abnormal behavior can be detected based on the vehicle speed that can be acquired according to the output of the vehicle speed sensor 3 or the like, the configuration of the abnormality detection system can be simplified. Further, the abnormal driving behavior detection device 1 can be applied regardless of whether it is a real vehicle or a virtual vehicle by a simulator. Therefore, the abnormal driving behavior detection device 1 not only performs abnormal notification to prevent a risk that may occur when driving on an actual road (actual driving), but also performs an aptitude test by actual driving or a simulator. It is also suitable for use. For example, it can be used for aptitude inspection by a driving school or periodic aptitude inspection (for example, aptitude inspection of taxis and truck drivers) in the transportation / transportation industry that requires a second-class motor driving license (business driving license).

(2) Specifically, the abnormal behavior detection unit 14 determines the vehicle speed (V (t)) acquired by the vehicle speed acquisition unit 12 at a time before the predetermined time and the vehicle speed at the predetermined time estimated by the vehicle speed estimation unit 13. Abnormal behavior is detected when the difference from (Var (t)) exceeds a predetermined threshold T2. Note that in realizing the real-time property of abnormality detection, the difference between the actual measured value V (t) of the vehicle speed acquired at time t and the estimated vehicle speed Var (t) at the next sample time of time t becomes the threshold value T2. It is more preferable that an abnormal behavior is detected when the number exceeds the limit.

(3) In addition, the order of the autoregressive model M is determined in advance, and the vehicle speed estimation means 13 adds the vehicle speed (V (t) · z ⁻ ) for the orders at different times before the predetermined time to the autoregressive model M. ¹ ,..., V (t) · z ⁻ⁿ ) are substituted to estimate the vehicle speed (Var (t)) at a predetermined time point.
(4) The autoregressive model M is determined in advance based on a time series of vehicle speeds when the vehicle 2 is traveled a plurality of times on a predetermined route in a state where no abnormal behavior occurs.
(5) Moreover, the abnormal driving action detection apparatus 1 is not restricted to what is mounted in the vehicle 2 as a real vehicle. The vehicle 2 is a virtual vehicle by a simulator and may be provided in the simulator.

(6) The abnormal driving behavior detection method using the abnormal driving behavior detection apparatus 1 described above includes an acquisition step of acquiring the vehicle speed of the vehicle 2 and a vehicle speed (Var (t)) at a predetermined time in advance. The determined autoregressive model M and a plurality of vehicle speeds (V (t), V (t) · z ⁻¹ ,..., V (t) · z ⁻ⁿ ) at different time points before a predetermined time point An abnormal behavior is detected when the difference between the estimation step for estimating the vehicle speed at a predetermined time based on the vehicle speed, the vehicle speed acquired at the acquisition step, and the vehicle speed at the predetermined time estimated at the estimation step exceeds a predetermined threshold T2. Detecting step.

(7) Moreover, the program P demonstrated above makes the computer (control part 10) determine beforehand the vehicle speed acquisition means 12 which acquires the vehicle speed of the vehicle 2, and the vehicle speed (Var (t)) at the predetermined time. And a plurality of vehicle speeds (V (t), V (t) · z ⁻¹ ,..., V (t) · z ⁻ⁿ ) at different time points before the predetermined time point. When the difference between the vehicle speed acquired by the vehicle speed estimation means 13 and the vehicle speed acquisition means 12 for estimating the vehicle speed at the predetermined time based on the vehicle speed at the predetermined time estimated by the vehicle speed estimation means 13 exceeds a predetermined threshold T2. It functions as an abnormal behavior detecting means 14 for detecting abnormal behavior.

  In the above description, in order to facilitate understanding of the present invention, descriptions of known technical matters are omitted as appropriate.

DESCRIPTION OF SYMBOLS 1 ... Abnormal driving action detection apparatus 10 ... Control part 11 ... Memory | storage part P ... Program, M ... Autoregressive model 12 ... Vehicle speed acquisition means 13 ... Vehicle speed estimation means 14 ... Abnormal action detection means 20 ... Notification part 2 ... Vehicle 3, 3 ... Vehicle speed sensor

Claims (7)

An abnormal driving behavior detection device that detects abnormal behavior related to acceleration / deceleration operations by a driver driving a vehicle,
Obtaining means for obtaining a vehicle speed of the vehicle;
Storage means for storing information indicating the vehicle speed acquired by the acquisition means and a predetermined autoregressive model for estimating the vehicle speed at a predetermined time;
Estimating means for estimating the vehicle speed at the predetermined time point based on the autoregressive model and a plurality of vehicle speeds at different time points before the predetermined time point;
Detecting means for detecting the abnormal behavior when a difference between the vehicle speed acquired by the acquisition means and the vehicle speed at the predetermined time estimated by the estimation means exceeds a predetermined threshold;
An abnormal driving behavior detecting device characterized by the above. The detecting means detects the abnormal behavior when the difference between the vehicle speed acquired by the acquiring means at a time before the predetermined time and the vehicle speed at the predetermined time estimated by the estimating means exceeds the threshold. To detect,
The abnormal driving behavior detecting device according to claim 1. The order of the autoregressive model is predetermined,
The estimating means substitutes the vehicle speed for the order at different times before the predetermined time into the autoregressive model to estimate the vehicle speed at the predetermined time.
The abnormal driving behavior detecting device according to claim 1 or 2, characterized in that. The autoregressive model is determined in advance based on a time series of vehicle speeds when the vehicle is caused to travel a plurality of times in a state where the abnormal behavior does not occur on the route determined in advance.
The abnormal driving behavior detection device according to any one of claims 1 to 3, wherein The vehicle is a virtual vehicle by a simulator, and is provided in the simulator.
The abnormal driving behavior detection device according to any one of claims 1 to 4, wherein the abnormal driving behavior detection device is provided. An abnormal driving behavior detection method for detecting abnormal behavior related to acceleration / deceleration operations by a driver driving a vehicle,
An acquisition step of acquiring a vehicle speed of the vehicle;
An estimation step for estimating the vehicle speed at the predetermined time point based on a predetermined autoregressive model for estimating the vehicle speed at the predetermined time point and a plurality of vehicle speeds at different time points before the predetermined time point;
A detection step of detecting the abnormal behavior when a difference between the vehicle speed acquired in the acquisition step and the vehicle speed at the predetermined time estimated in the estimation step exceeds a predetermined threshold;
An abnormal driving behavior detection method characterized by the above. A program for detecting abnormal behavior related to acceleration / deceleration operations by a driver driving a vehicle,
Computer
Obtaining means for obtaining the vehicle speed of the vehicle;
Estimating means for estimating the vehicle speed at the predetermined time point based on a predetermined autoregressive model for estimating the vehicle speed at the predetermined time point and a plurality of vehicle speeds at different time points before the predetermined time point;
Detecting means for detecting the abnormal behavior when the difference between the vehicle speed acquired by the acquiring means and the vehicle speed at the predetermined time estimated by the estimating means exceeds a predetermined threshold;
Program to function as.