KR20200018838A - System and method for representing dangerous driving - Google Patents

Abstract

The present invention relates to a system and a method for preventing dangerous driving due to drowsiness or carelessness of a driver. The system for preventing dangerous driving according to the present invention includes: a driver information acquisition module that acquires driver information; and an ECU module that determines driver′s conditions using the driver information, vehicle driving information, and a pre-trained classification model, and provides warnings for careless driving or drowsy driving.

Description

Dangerous driving prevention system and its method {SYSTEM AND METHOD FOR REPRESENTING DANGEROUS DRIVING}

The present invention relates to a system and method for preventing dangerous driving due to drowsiness or carelessness of a driver.

According to the National Highway Safety Association (NTHSA), more than 1.2 million people die every year from traffic accidents worldwide, of which 94% of all accidents are caused by human error.

Human errors include speeding, violent driving, drunk driving, careless driving, and drowsy driving.

According to the prior art, in order to determine the driver's carelessness or drowsy driving, the driver's head movement, blinking eyes, vital signs, and the like are used.

However, vibrations inevitably occurring in the driving environment, not really due to inattention or drowsiness, but similar driver movements (for example, when looking at the front side rather than the road ahead in a left / right turn situation, or talking with a passenger) The recognition rate is lowered or a misrecognition occurs due to the case of nodding, etc.), thereby providing an unnecessary warning to the driver while driving and thus failing to secure the reliability and, as a result, the technology cannot be commercialized. .

The present invention has been proposed to solve the above-described problems, using the driver's state information and the driving information of the vehicle, the normal state, careless state, drowsiness state, exception state (when turning left / right, lane change, etc.) The dangerous driving prevention system capable of minimizing false recognition and unrecognized recognition of the driver's status and increasing the reliability of the warning of the dangerous situation by determining the driver's condition based on the state in which the warning of the dangerous driving should be deactivated). The purpose is to provide a method.

The dangerous driving prevention system according to the present invention determines a driver's condition by using a driver information acquisition module for acquiring driver information, driver information, vehicle driving information, and a pre-learned classification model, and provides a warning about careless driving or drowsy driving. It characterized in that it comprises an ECU module.

According to an embodiment of the present invention, when compared to the driver state determination technology according to the prior art using a single factor, such as head movement, opening and closing of eyes, and a biosignal through a wearable band, a plurality of driver information may be converged. Judgment has the effect of improving the accuracy of judgment.

By combining driver information and vehicle driving information, the driver's condition is classified, and whether or not dangerous driving (inadvertent driving or drowsy driving) is provided and warning is provided, thereby increasing the reliability of providing the warning and thus increasing the driver's acceptance. It works.

The effects of the present invention are not limited to those mentioned above, and other effects that are not mentioned will be clearly understood by those skilled in the art from the following description.

1 is a block diagram illustrating a dangerous driving prevention system according to an exemplary embodiment of the present invention.
2 is an exemplary view showing a driver information obtaining module of a glasses type according to an embodiment of the present invention.
3 is a block diagram illustrating a CPU of an ECU module according to an embodiment of the present invention.
4 is a flowchart illustrating a dangerous driving prevention method according to an exemplary embodiment of the present invention.

BRIEF DESCRIPTION OF THE DRAWINGS The above and other objects, advantages and features of the present invention, and methods of achieving them will be apparent with reference to the embodiments described below in detail with the accompanying drawings.

However, the present invention is not limited to the embodiments disclosed below, but may be implemented in various forms, and only the following embodiments are provided to those skilled in the art to which the present invention pertains. It is merely provided to easily show the configuration and effects, the scope of the invention is defined by the claims.

Meanwhile, the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. In this specification, the singular also includes the plural unless specifically stated otherwise in the phrase. As used herein, “comprises” and / or “comprising” refers to the presence of one or more other components, steps, operations and / or devices in which the mentioned components, steps, operations and / or devices are known. Or does not exclude addition.

1 is a block diagram illustrating a dangerous driving prevention system according to an exemplary embodiment of the present invention.

The dangerous driving prevention system according to an exemplary embodiment of the present invention determines a driver's state by using the driver information acquisition module 100 for acquiring the driver information, the driver information, the vehicle driving information, and the pre-learned classification model, and careless driving or drowsiness. ECU module 200 that provides a warning about operation.

As shown in FIG. 2, the driver information acquisition module 100 is preferably a type of glasses worn by a driver. In this case, the first gyro sensor 110, the camera 121: an RGB camera, and the 122: an infrared camera. The driver's face motion information, pulse wave information, and opening / closing / eye size information of the driver are acquired by using the glasses type driver information acquisition module.

In another embodiment, the first gyro sensor may be disposed on one side of the glasses worn by the driver, and the camera may be separately disposed in the vehicle to obtain driver's face movement information, pulse wave information, and eye information, and heart rate information may be separately provided. It is also possible to obtain from a wearable device.

Referring to FIG. 1, the driver information acquisition module 100 may include a first gyro sensor 110, a camera 120, and a first communication unit 130, and the first gyro sensor 110 may have a horizontal direction and a vertical direction. Sensing the driver's face movement with respect to, the camera 120 senses the opening and closing or size of the driver's eyes, pulse wave information, the first communication unit 130 transmits the sensed driver information to the ECU module 200.

Preferably, the first communication unit 130 transmits the sensed information through Bluetooth, and transmits the sensed information to the paired PC, smartphone, etc. as well as the second communication unit 230 of the ECU module 200. Do.

The ECU module 200 inputs a second gyro sensor 210 for acquiring vehicle movement information, driver information, and vehicle driving information to a pre-learned classification model to determine whether a dangerous situation and a driver state are present. , A second communication unit 230.

The second communication unit 230 receives driver information transmitted through the first communication unit 130, and transmits a warning command signal to the driver information obtaining module 100, the vehicle device, or an external server.

The driver information obtaining module 100 and the vehicle device, which have received the warning command signal from the second communication unit 230, provide the driver with warning about careless driving or drowsy driving through vibration, sound, and warning messages.

The second communication unit 230 transmits a warning command signal to the autonomous driving system of the driving vehicle, so that autonomous driving can be performed by transferring control of the driver from the driver (autonomous driving to the destination, or inattention or drowsiness of the driver). Autonomous driving in various modes is possible, such as autonomous driving until it is determined that driving is resolved, or evacuation of a vehicle to sleepy shelter, or by sending a warning command signal to an external server for emergency situations (e.g. It is also possible to request rescue assistance for situations, etc.).

The second gyro sensor 210 senses the movement of the vehicle, and the ECU module 200 detects the movement information of the vehicle acquired through the second gyro sensor 210 and the face of the driver acquired by the first gyro sensor 110. By comparing the motion information, the driver's movement in the vehicle is identified.

The CPU 220 of the ECU module 200 uses the IVN information of the vehicle received through the CAN transceiver and the driver information received from the driver information acquisition module 100, through a classification model previously learned. Determines whether the driver is inadvertent or drowsy driving.

According to an embodiment of the present invention, the ECU module 200 minimizes misperception and unrecognition of driver's state determination by fusing driver's face movement, eye open / closed state and eye size information, and heart rate information instead of sole factor. .

[Table 1] below shows the determination accuracy when the single factor is used to determine whether drowsy driving is determined and when the convergence of various driver information and vehicle information is performed.

Factor Vision Bio-Signal Vehicle data Vision & Bio Vision & vehicle Vehicle & Bio All Classification Model Accuracy Decision tree 82.8 87.0 81.2 94.6 94.5 89.4 96.5 SVM 80.7 62.8 67.5 86.5 82.7 68.1 86.7 KNN 83.5 88.3 88.4 84.8 94.5 91.2 95.4

ECU module 200 according to an embodiment of the present invention by combining the driving information of the vehicle and the driver information received from the driver information acquisition module 100 to determine whether the dangerous situation and the driver state, if not driving, left / In case of a right turn, in case of a lane change, etc., the warning is deactivated.

In addition, the driver information and the ADAS information (distance from the front vehicle, the lateral position in the lane, etc.) are used to determine whether inadvertent driving or drowsy driving, and to provide a warning about it.

Hereinafter, with reference to FIG. 2, a driver information obtaining module of a glasses type according to an embodiment of the present invention will be described.

The driver information acquisition module 100 includes a first gyro sensor 110, an RGB camera 121, an infrared camera 122, a first communication unit 130, a first CPU 140, and a battery 150.

The infrared camera 122 is installed at the lower portion of the spectacle lens in order to minimize the occlusion of the driver's field of view, and acquires information regarding the opening and closing of the driver's eyes or the size of the eyes.

The first gyro sensor 110 senses horizontal and vertical movements of the driver's face.

The RGB camera 121 senses pulse wave information of the driver.

The first CPU 140 obtains driver information from the first gyro sensor 110, the RGB camera 121, and the infrared camera 122, and the first communication unit 130 obtains the driver information using the Bluetooth method as described above. Transmit to ECU module 200.

Hereinafter, the second CPU of the ECU module according to the embodiment of the present invention will be described with reference to FIG. 3.

The second CPU 220 inputs an input unit 221 for receiving driver information and vehicle driving information, and a dangerous driving determination unit for determining whether a dangerous situation and a driver state are input by inputting the driver information and vehicle driving information to a previously learned classification model. 222 and an output unit 223 for transmitting a warning providing command signal.

The input unit 221 receives the driver information acquired by the driver information acquisition module 100 and the vehicle driving information received through the CAN transceiver and transmits them to the dangerous driving determination unit 222.

The dangerous driving determination unit 222 includes a dangerous situation determination unit 222-1 for determining whether or not a dangerous situation, and a driver state determination unit 222-2 for classifying and determining a driver state in a dangerous situation.

The dangerous driving determination unit 222 may determine whether a dangerous situation is performed by using the face movement information of the driver included in the driver information and the vehicle movement information included in the vehicle driving information, for example, through the first gyro sensor 110. Even if the acquired driver's face movement deviates to the left by more than a predetermined angle (for example, 30 degrees) based on the forward basis, the movement information (compare target information) of the driving vehicle acquired through the second gyro sensor 210 is turned left or In the case of making a U-turn to the left, it is determined as a normal state without discriminating it as a careless state of the driver.

The dangerous driving determination unit 222 acquires sound information in the vehicle through a separate device disposed in the vehicle or a microphone disposed in the driver information acquisition module 100, and obtains the sound information from the vehicle through the first gyro sensor 110. The driver's state is determined by fusion with the head movement information of the camera.

For example, when the driver nods his head as a sign of consent or talks with his passenger, he may be mistaken for drowsy driving.

That is, when the head motion information of the driver acquired through the first gyro sensor 110 is used as a single factor for the determination of drowsiness driving, it may be mistaken as drowsy driving, and thus dangerous driving according to an embodiment of the present invention. The determination unit 222 does not rely only on the head movement information of the driver acquired through the first gyro sensor 110, and inputs it to the classification model together with the vehicle sound information, which is the comparison target information, to prevent misrecognition. .

The classification model according to an embodiment of the present invention includes a plurality of tree structures according to random forest learning. The dangerous driving determination unit 222 may use a bagging tree-based classification model using driver information and vehicle driving information as input variables. The driver's state is determined by using the normal state, the careless state, the drowsiness state, and the exception state.

The dangerous driving determination unit 222 calculates input data by performing preprocessing on the above-described driver information and vehicle driving information.

For example, the closed data is processed to calculate the closed time and the closed time per minute as input data, which is input to the classification model.

Table 2 below lists the variables input to the bagging tree-based classification model according to an embodiment of the present invention.

No. Input variables Explanation unit Acquisition module Remarks One EyeLidOpening Eyes closed 0: Unrecognized, 1: Open, 2: Close infrared ray
camera - 2 Head_Direction_Horizontal_deg Face horizontal movement Deg Gyro sensor - 3 Heartrate Heart rate Times / min RGB camera - 4 Velocity speed Kph vehicle - 5 APS Excel pedal
Degree % vehicle - 6 Engine_RPM Rpm 7 Brake_SW 0: off, 1: on 8 Long_accel Longitudinal acceleration m / s2 vehicle - 9 Lat_accel Lateral acceleration m / s2 vehicle - 10 Turnsignal Turn indicator 0: off, 1: on vehicle - 11 ClacClosedTimes Closed time sec - 1 data
Processing using 12 ClacPerclos Per minute
Closed time - - 1 data
Processing using 13 ClacDistractionTimes Forward
Breakaway time sec - Data 2
Processing using 14 ZigzagDetection Zigzag operation detection 0: not detected, 1: detected - Processing using the distance from the lane acquired through the vehicle's ADAS camera 15 SafeDistance Safety distance not detected 0: not detected, 1: detected - Processing using distance from front vehicle acquired through ADAS camera of vehicle 16 FCWarning_ECU Forward collision warning 0: no warning
1: warning vehicle 17 LDWarning_ECU Lane Departure Warning 0: no warning1: warning vehicle 18 HR_ClacVariable1 Heart rate variability - - Data 3
Processing using

Bagging, which stands for bootstrap aggregating, is a method of aggregating trained base learners for slightly different training data through bootstrap.

When classification target data is input to a classification model according to an exemplary embodiment of the present invention, a determination result is inputted to a classifier of each tree to calculate a determination result, and a final determination of a driver's state is performed through voting using each determination result.

According to an exemplary embodiment of the present invention, the result of the determination of the dangerous driving determination unit 222 may be 0 (normal state), 1 (careless state), 2 (sleepiness state), 3 (exception state: left / right turn or ramp section). This can be

According to an embodiment of the present invention, the bagging treee has a predetermined number of tree structures (for example, 30), one of which is shown in Table 3 below.

1 if LDWarning_ECU <2 then node 2 elseif LDWarning_ECU> = 2 then node 3 else 0
2 if Long_Accel <5.82231 then node 4 elseif Long_Accel> = 5.82231 then node 5 else 0
3 if HeartRate <88.5 then node 6 elseif HeartRate> = 88.5 then node 7 else 3
4 if ClacPerclos <0.369167 then node 8 elseif ClacPerclos> = 0.369167 then node 9 else 0
5 if Velocity <60.5 then node 10 elseif Velocity> = 60.5 then node 11 else 0
6 if Velocity <59.5 then node 12 elseif Velocity> = 59.5 then node 13 else 3
7 if ClacPerclos <0.558333 then node 14 elseif ClacPerclos> = 0.558333 then node 15 else 0
8 if ClacPerclos <0.220833 then node 16 elseif ClacPerclos> = 0.220833 then node 17 else 0
9 if ClacClosedTimes <0.04585 then node 18 elseif ClacClosedTimes> = 0.04585 then node 19 else 0
10 if EyeLidOpening <1.5 then node 20 elseif EyeLidOpening> = 1.5 then node 21 else 0
11 if EyeLidOpening <1.5 then node 22 elseif EyeLidOpening> = 1.5 then node 23 else 0
12 if Engine_RPM <1223.5 then node 24 elseif Engine_RPM> = 1223.5 then node 25 else 0
13 if HR_ClacVariable1 <5.9691 then node 26 elseif HR_ClacVariable1> = 5.9691 then node 27 else 3
14 if Engine_RPM <1260 then node 28 elseif Engine_RPM> = 1260 then node 29 else 0
15 if HR_ClacVariable1 <7.05243 then node 30 elseif HR_ClacVariable1> = 7.05243 then node 31 else 2
16 if Engine_RPM <275 then node 32 elseif Engine_RPM> = 275 then node 33 else 0
17 if Engine_RPM <1567 then node 34 elseif Engine_RPM> = 1567 then node 35 else 0
18 if FCWarning_ECU <0.5 then node 36 elseif FCWarning_ECU> = 0.5 then node 37 else 0
19 if Head_Direction_Horizontal_deg <-0.5 then node 38 elseif Head_Direction_Horizontal_deg> =-0.5 then node 39 else 0
20 if Engine_RPM <806.5 then node 40 elseif Engine_RPM> = 806.5 then node 41 else 0
21 if Velocity <38.5 then node 42 elseif Velocity> = 38.5 then node 43 else 0
22 if Head_Direction_Horizontal_deg <-6.5 then node 44 elseif Head_Direction_Horizontal_deg> =-6.5 then node 45 else 0
23 if ClacPerclos <0.300833 then node 46 elseif ClacPerclos> = 0.300833 then node 47 else 0
24 if HeartRate <55.5 then node 48 elseif HeartRate> = 55.5 then node 49 else 0
25 if APS <47 then node 50 elseif APS> = 47 then node 51 else 2
26 if TurnSignal <0.5 then node 52 elseif TurnSignal> = 0.5 then node 53 else 3
27 if Head_Direction_Horizontal_deg <-69.5 then node 54 elseif Head_Direction_Horizontal_deg> =-69.5 then node 55 else 3
28 if HeartRate <90.5 then node 56 elseif HeartRate> = 90.5 then node 57 else 2
29 if APS <19 then node 58 elseif APS> = 19 then node 59 else 0
...
69679 if HeartRate <69.5 then node 69696 elseif HeartRate> = 69.5 then node 69697 else 3
69680 class = 0
69681 class = 2
69682 if Head_Direction_Horizontal_deg <-1 then node 69698 elseif Head_Direction_Horizontal_deg> =-1 then node 69699 else 3
69683 class = 3
69684 class = 0
69685 class = 3
69686 class = 3
69687 if Engine_RPM <1284.5 then node 69700 elseif Engine_RPM> = 1284.5 then node 69701 else 2
69688 class = 0
69689 class = 3
69690 class = 3
69691 class = 0
69692 if ClacPerclos <0.233333 then node 69702 elseif ClacPerclos> = 0.233333 then node 69703 else 2
69693 if ClacPerclos <0.2275 then node 69704 elseif ClacPerclos> = 0.2275 then node 69705 else 0
69694 class = 0
69695 class = 2
69696 class = 0
69697 class = 3
69698 class = 3
69699 if ClacPerclos <0.314167 then node 69706 elseif ClacPerclos> = 0.314167 then node 69707 else 0
69700 class = 2
69701 if Velocity <71.5 then node 69708 elseif Velocity> = 71.5 then node 69709 else 2
69702 class = 0
69703 class = 2
69704 class = 0
69705 if Lat_Accel <155.364 then node 69710 elseif Lat_Accel> = 155.364 then node 69711 else 0
69706 class = 0
69707 class = 3
69708 class = 0
69709 if Head_Direction_Horizontal_deg <-1 then node 69712 elseif Head_Direction_Horizontal_deg> =-1 then node 69713 else 2
69710 class = 0
69711 class = 2
69712 class = 2
69713 if HR_ClacVariable1 <1.38248 then node 69714 elseif HR_ClacVariable1> = 1.38248 then node 69715 else 2
69714 class = 3
69715 class = 2

The dangerous driving determination unit 222 according to the exemplary embodiment of the present invention performs the determination on the final driver state by a vote method through the result output from the predetermined number of Tree structures as described above.

For example, when the speed of the traveling vehicle is 60 kph and the driver keeps his eyes closed, if ClacClosedTimes is 5 seconds or more, it is determined that he is drowsy.

In the same vehicle speed, if ClacPerclos is more than 0.4 and HR_ClacVariable1 is more than 10, it is judged as drowsiness. In addition, if the ClacDistractionTimes is 3 seconds or more in a state where Head_Direction_Horizontal_deg is 30 or more at the same vehicle speed, it is determined that the vehicle is in a careless state.

The dangerous driving determination unit 222 according to an embodiment of the present invention is a tree structure included in the classification model in consideration of driving conditions of various vehicles such as road oscillation magnitude, day / night driving time zone, tunnel driving situation, etc. on various roads. Select and adopt tree for driver status judgment.

In order to determine driver status information of a traveling vehicle and provide a warning according to the situation, it is important to secure not only a short calculation speed but also a reliable judgment result.

Accordingly, in consideration of the driving environment, it is possible to exclude the tree structure predicted to be unreliable and to perform the driver state determination of the remaining tree structure, thereby improving not only the operation speed but also the reliability of the determination result.

In another exemplary embodiment, the dangerous driving determination unit 222 may give different weights to vote results calculated from each tree structure in consideration of driving environment information in finally determining a driver's state.

For example, when the road surface vibration is severe, a relatively low weight is given to a result calculated by a tree structure having the driver's face motion information obtained using the first gyro sensor 110 as an input variable.

In other words, in performing the final judgment on the driver's condition, the result calculated from each tree structure is not counted with the same weight, but the weight of the tree structure that is expected to be highly reliable in the driving environment is relatively high. And by applying this to the final judgment, it is effective to increase the reliability of the final judgment result for the driver state calculated by the vote method.

4 is a flowchart illustrating a dangerous driving prevention method according to an exemplary embodiment of the present invention.

According to an embodiment of the present invention, a method for preventing dangerous driving includes obtaining driver information and driving information of a vehicle (S100), using driver information, driving information of a vehicle, and a classification model that has been learned. Calculating a state determination result (S200); and providing a warning (S300) when the state determination result of the driver corresponds to careless driving or drowsy driving.

In operation S100, driver information including at least one of driver's face movement information, eye closure information, and heart rate information is acquired from the driver information acquisition module.

As described above, the driver information acquisition module may be a glasses type, and may acquire driver information including a first gyro sensor, an infrared camera, and an RGB camera. One gyro sensor is arranged, the eye information and heart rate information of the driver can also be obtained from a camera or a separate wearable device that is disposed separately in the vehicle.

Step S200 determines whether the driver's face movement is a warning target by using the movement information of the vehicle or the sound information in the vehicle.

The driver's nod may be made not only in the case of drowsy driving, but also when talking and agreeing with a passenger, or playing music and nodding accordingly.

Therefore, in step S200, the motion information of the driver's head obtained through the first gyro sensor is obtained by using the vehicle's motion information obtained through the second gyro sensor or the in-car sound information obtained through a separate microphone as the comparison target information. Determine whether this is a significant factor for determining drowsy driving.

In step S200, the driver's state is determined based on the classification of the normal state, the careless state, the drowsiness state, and the exception state using a classification model including a plurality of tree structures according to random forest learning.

In step S200, the driver state is determined using a classification model based on bagging tree using the driver information and vehicle driving information as input variables, and at this time, the reliability of the driving condition among the trees included in the classification model is considered in consideration of the driving conditions of the vehicle. By selecting and adopting the tree that is expected to be high, or by assigning different weights to the vote result calculated from each tree structure in consideration of the driving situation and performing the final judgment, it is possible to improve the computation speed and secure the reliability of the judgment result. have.

The step S300 is to provide a warning when the driver's condition is inadvertent driving or drowsy driving, and the warning is provided through a speaker or a vibrating body disposed in the driver information acquisition module, or by utilizing a device in the vehicle, warning sounds, seat vibrations, and warnings. It is possible to be provided in a manner such as a message.

The step S300 provides a warning to the driver, and also transmits a request signal to the autonomous driving system of the vehicle to request the transfer of driving control, so that the vehicle can be converted to autonomous driving from a careless or drowsy driving state.

So far looked at the center of the embodiments of the present invention. Those skilled in the art will appreciate that the present invention can be implemented in a modified form without departing from the essential features of the present invention. Therefore, the disclosed embodiments should be considered in descriptive sense only and not for purposes of limitation. The scope of the present invention is shown in the claims rather than the foregoing description, and all differences within the scope will be construed as being included in the present invention.

100: driver information acquisition module 110: first gyro sensor
120: camera 121: RGB camera
122: infrared camera 130: first communication unit
140: first CPU 150: battery
200: ECU module 210: second gyro sensor
220: second CPU 221: input unit
222: Dangerous driving judgment unit 222-1: Dangerous situation determination unit
222-2: driver state determination unit 223: output unit
230: second communication unit

Claims (16)

A driver information obtaining module for obtaining driver information; And
ECU module for determining the driver's condition by using the driver information, vehicle driving information and the pre-learned classification model and providing a warning about careless driving or drowsy driving
Dangerous driving prevention system comprising a.
The method of claim 1,
The driver information obtaining module obtains at least one of face motion information, eye closure information, and heart rate information and transmits it to the ECU module.
Dangerous driving prevention system.
The method of claim 1,
The ECU module inputs the driver information and vehicle driving information to the previously learned classification model, wherein the previously learned classification model includes a plurality of tree structures according to random forest learning.
Dangerous driving prevention system.
The method of claim 3,
The ECU module determines the driver state by using a bagging tree-based classification model using the driver information and vehicle driving information as input variables, and classifies the driver state into a normal state, a careless state, a drowsy state, and an exception state. Judging
Dangerous driving prevention system.
The method of claim 3,
The ECU module determines whether the head motion information is a warning object by using comparison object information and head motion information of a driver acquired using a first gyro sensor included in the driver information acquisition module, and compares the head motion information with a warning object. The target information is motion information of the vehicle acquired using the second gyro sensor or sound information of the vehicle.
Dangerous driving prevention system.
The method of claim 3,
The ECU module selects and adopts a tree structure included in the previously learned classification model in consideration of driving environment information.
Dangerous driving prevention system.
(a) obtaining driver information and driving information of the vehicle;
(b) calculating a risk situation and a driver's state determination result using the driver information, driving information of the vehicle, and a previously learned classification model; And
(c) providing a warning if the driver's state determination result corresponds to inadvertent driving or drowsy driving;
Dangerous driving prevention method comprising a.
The method of claim 7, wherein
In the step (a), obtaining the driver information including at least one of a driver's face movement information, eye closure information, and heart rate information.
To prevent dangerous driving.
The method of claim 7, wherein
In the step (b), it is determined whether the driver's face movement is a warning target by using the movement information of the vehicle or the sound information of the vehicle.
To prevent dangerous driving.
The method of claim 7, wherein
Step (b) is to determine the driver's state with respect to the classification of steady state, careless state, drowsiness state and exception state using the classification model including a plurality of tree structures according to random forest learning.
To prevent dangerous driving.
The method of claim 10,
In step (b), the logic included in the previously learned classification model is selected in consideration of driving environment information, and the result of determining the state of the driver is calculated.
To prevent dangerous driving.
The method of claim 7, wherein
In step (c), the warning is provided through a device in a vehicle or a wearable device of a driver.
To prevent dangerous driving.
An input unit to receive driver information and vehicle driving information;
A dangerous driving determination unit configured to input the driver information and the vehicle driving information into a classification model learned in advance to determine whether a dangerous situation and a driver state are present; And
Output unit for transmitting a warning providing command signal according to the determination result of the dangerous situation and the driver status
Dangerous driving prevention ECU module comprising a.
The method of claim 13,
The dangerous driving determination unit determines whether the dangerous situation using the face movement information of the driver included in the driver information and the vehicle movement information included in the vehicle driving information.
Dangerous driving prevention ECU module.
The method of claim 13,
The dangerous driving determination unit determines a driver state classified into a normal state, a careless state, a drowsiness state, and an exception state using a bagging tree based classification model using the driver information and vehicle driving information as input variables.
Dangerous driving prevention ECU module.
The method of claim 13,
The dangerous driving determination unit selects and adopts a tree for determining a driver's state among tree structures included in the classification model in consideration of driving environment information.
Dangerous driving prevention ECU module.

Images