Danger driving warning device, vehicle, and danger driving warning method

Abstract

To provide a danger driving warning device capable of urging safety driving to a driver with warning contents corresponding to a predicted danger level by calculating a danger driving occurrence probability based on visual line information of the driver.SOLUTION: A danger driving warning device comprises: a map acquisition section for acquiring map information including obstacle positions; a visual line information acquisition section for acquiring visual line information of a driver who drives a vehicle; a vehicle information acquisition section for acquiring vehicle information including a vehicle speed and a steering angle; a visible probability calculation section for calculating a visible probability of the driver in a visible area of the driver with the use of the visual line information; a vehicle trajectory calculation section for calculating a vehicle trajectory with the use of the vehicle information; a danger level calculation section for calculating a danger level being the probability of danger driving occurrence with the use of the map information, the visible probability, and the vehicle trajectory; and an output section for outputting warning to the driver when the danger level is equal to or more than an arbitrary threshold.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a dangerous driving warning device that predicts the occurrence of dangerous driving and warns a driver, a vehicle equipped with the device, and a dangerous driving warning method. [Background technology]

[0002] In recent years, the development of advanced driver-assistance systems (hereinafter referred to as "ADAS") installed in automobiles has progressed rapidly, and one of the functions proposed for such systems is a function to predict dangerous driving and warn the driver. As a method for predicting dangerous driving in such a system, there is a method that uses a judgment based on the driver's line of sight information. As a driving assistance system that takes line of sight information into consideration, the driving assistance system of Patent Document 1 is known.

[0003] The abstract of Patent Document 1 states that the problem is to "enable appropriate driving assistance." and states that the solution is to "provide appropriate driving assistance." The abstract states that "the first imaging unit 201 images the outside of the vehicle, and the second imaging unit 202 images the driver. The image processing unit 203 detects the driver's line of sight. The information acquisition unit 205 acquires situation information indicating the surrounding environment. The driving parameter information generation unit 206 generates driving parameter information using the current position detected by the self-position detection unit 204, the driving control information generated by the driving control unit, information indicating the driver's line of sight, information indicating the surrounding environment, etc. as parameter information. The judgment unit 209 performs risk judgment based on the difference between the driving parameter information indicating the state of the vehicle or the driver, and optimal parameter information indicating the optimal state of the vehicle or the driver in the current position and the current surrounding environment. The information presentation unit 214 presents driving assistance information based on the judgment result, and the driving control unit 210 performs driving assistance based on the judgment result, thereby reducing the difference in parameter information."

[0004] In this way, the driving assistance system of Patent Document 1 judges danger based on the difference between current driving parameter information (driver's gaze information, as well as the vehicle's current position, driving control information, surrounding environment information, etc.) and optimal parameter information. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Patent Publication No. 2021-006967 Summary of the Invention [Problem to be solved by the invention]

[0006] However, the driving assistance system of Patent Document 1 only distinguishes whether or not there is danger based on gaze information (driving parameter information) at a specific time (see ST7 in Figure 4, paragraph 0052 of the same document, etc.), and is unable to more accurately predict the probability of dangerous driving occurring (hereinafter also referred to as the "danger level") by taking into account changes in gaze information over time, or to encourage the driver to drive safely with a warning content according to the predicted danger level.

[0007] Therefore, the present invention aims to provide a dangerous driving warning device, vehicle, and dangerous driving warning method that can predict a danger level based on time-series changes in a driver's gaze information and encourage the driver to drive safely with warning content that corresponds to the predicted danger level. [Means for solving the problem]

[0008] In order to solve the above problems, one of the dangerous driving warning devices of the present invention includes a map acquisition unit that acquires map information including the position of an obstacle, a gaze information acquisition unit that acquires gaze information of the driver driving the vehicle, a vehicle information acquisition unit that acquires vehicle information including the vehicle speed and steering angle, a visibility probability calculation unit that uses the gaze information to calculate the visibility probability of the driver in an area visible to the driver, a vehicle trajectory calculation unit that uses the vehicle information to calculate the trajectory of the vehicle, a danger level calculation unit that calculates a danger level, which is the probability that dangerous driving will occur, using the map information, the visibility probability, and the vehicle trajectory, and an output unit that outputs a warning to the driver if the danger level is equal to or above an arbitrary threshold value. Effect of the Invention

[0009] The dangerous driving warning device, vehicle, and dangerous driving warning method of the present invention can estimate the danger level based on time-series changes in the driver's gaze information, and encourage the driver to drive safely with warning content that corresponds to the predicted danger level. [Brief description of the drawings]

[0010] [Figure 1] FIG. 2 is a functional block diagram of the dangerous driving warning device according to the first embodiment. [Diagram 2] 3 is a process flowchart of the dangerous driving warning device according to the first embodiment. [Diagram 3] FIG. 11 is a top view showing an example of a time series change in gaze information. [Figure 4] FIG. 4 is a diagram for explaining a time series change in the visibility probability of a pedestrian in the driving environment of FIG. 3. [Diagram 5] FIG. 11 is a top view illustrating an example of a situation in which dangerous driving occurs. [Figure 6] FIG. 11 is a top view illustrating an example of a situation in which dangerous driving occurs. [Figure 7] FIG. 11 is a top view illustrating an example of a situation in which dangerous driving occurs. [Figure 8] FIG. 8 is a diagram for explaining an example of a setting state of a safety region in FIG. 7; [Figure 9] FIG. 9 is a diagram for explaining an example of transition of the risky driving occurrence probability in FIG. 8 . [Figure 10] FIG. 6 is a diagram for explaining a method of notifying the occurrence of dangerous driving in FIG. 5; [Figure 11] FIG. 11 is a functional block diagram of a dangerous driving warning device according to a second embodiment. [Figure 12] 10 is a process flowchart of a dangerous driving warning device according to a second embodiment. [Figure 13] FIG. 13 is a diagram for explaining an example of creating a cognitive grid map. [Figure 14] FIG. 13 is a diagram illustrating a safety grid map in which an occupancy grid map and a perception grid map are superimposed. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an embodiment of a dangerous driving warning device according to the present invention will be described with reference to the drawings. EXAMPLES

[0012] First, a dangerous driving warning device 1 according to a first embodiment of the present invention will be described with reference to Fig. 1 to Fig. 10. In the following description, the dangerous driving warning device 1 will be described as being part of an in-vehicle driving assistance system 100, but the dangerous driving warning device 1 may be part of a driving simulator.

[0013] The driving assistance system 100 of this embodiment has an external environment recognition sensor 2, a vehicle sensor 3, and an HMI device 4 (Human Machine Interface) in addition to the dangerous driving warning device 1 of FIG. 1, and is mounted on a vehicle (hereinafter referred to as "vehicle V"). The external environment recognition sensor 2 is a sensor capable of recognizing the surrounding environment of the vehicle V, such as a monocular camera, a stereo camera, a laser radar, a millimeter wave radar, an ultrasonic sensor, an infrared sensor, etc. The vehicle sensor 3 is a sensor capable of acquiring the position information, longitudinal acceleration, yaw rate, etc. of the vehicle V, such as a wheel speed sensor, an acceleration sensor, an angular velocity sensor, a terminal of a global navigation satellite system (GNSS), etc. The HMI device 4 is a device for interaction with the driver, such as an eye tracker or an in-vehicle camera that acquires the driver's line of sight, etc., a touch-type display that can be operated by the driver, a speaker that issues a warning to the driver, etc. The GNSS is a device that provides various information about the road on which the vehicle V is traveling (such as the position information of stationary obstacles, traffic rules, etc.), and is a so-called car navigation system.

[0014] 1 is a functional block diagram of a dangerous driving warning device 1. As shown in this figure, the dangerous driving warning device 1 of this embodiment has functional units, namely, a map acquisition unit 11, a line-of-sight information acquisition unit 12, a vehicle information acquisition unit 13, a visibility probability calculation unit 14, a vehicle trajectory calculation unit 15, a danger level calculation unit 16, and an output unit 17, and is a device that generates a warning to a driver based on map information I1, line-of-sight information I2, and vehicle information I3 input from the outside.

[0015] Specifically, the dangerous driving warning device 1 is a computer equipped with hardware such as a calculation device such as a CPU (Central Processing Unit) and a storage device such as a semiconductor memory, and the calculation device executes a dangerous driving warning program stored in the storage device to realize each of the above-mentioned functional units. Some or all of this hardware may be replaced with a dedicated device, a general-purpose machine learning machine, a DSP (Digital Signal Processor), an FPGA (Field-Programmable Gate Array), a GPU (Graphics Processing Unit), a PLD (Programmable Logic Device), or the like.

[0016] Next, a warning generation process performed by the dangerous driving warning device 1 will be described with reference to the flowchart of FIG.

[0017] <Step S1> First, in step S1, the map acquisition unit 11 acquires map information I1 from the external recognition sensor 2 or the vehicle sensor 3 (GNSS). The acquired map information I1 is mainly information that indicates the position of an obstacle in a coordinate system centered on the vehicle or a fixed coordinate system, and includes position information of stationary obstacles and position information of stationary obstacles and moving obstacles detected by the external recognition sensor 2. Note that an occupancy grid map that is widely used in the fields of autonomous driving and robotics may be used as the map information I1 as it is. The map information I1 may also include information on traffic rules indicated by signs, road paint, traffic lights, etc. detected by the external recognition sensor 2.

[0018] <Step S2> Next, in step S2, the gaze information acquisition unit 12 acquires the driver's gaze information I2 from the HMI device 4. Here, the gaze information I2 is information indicating the direction in which the driver is actually looking with his / her eyes. This gaze information I2 may be acquired directly by an eye tracker worn by the driver, or may be estimated using an image captured by an in-vehicle camera. Moreover, the gaze information I2 may be estimated from the driver's gaze information and head-swivel angle.

[0019] 3 is a top view showing an example of a time series change in the gaze information I2 when the vehicle V is traveling on a straight road. The gaze information I2 can be expressed as a linear gaze L, but since the peripheral area of ​​the gaze L can also be visually observed, it may be expressed as a substantially sector-shaped field of view F including the peripheral area of ​​the gaze L. Furthermore, if the gaze L or field of view F includes a side mirror or a rearview mirror, the gaze L or field of view F may be expanded to the areas to the sides or rear of the vehicle.

[0020] <Step S3> In step S3, the vehicle information acquisition unit 13 acquires vehicle information I3 from the vehicle sensor 3. The vehicle information I3 includes, but is not limited to, for example, the vehicle position, vehicle speed, longitudinal acceleration, lateral acceleration, yaw rate, steering angle, steering angular velocity, etc. Note that the longitudinal acceleration can be calculated from the time derivative of the vehicle speed, and the lateral acceleration can be calculated from the relationship between the yaw rate and the speed, and it is not necessarily necessary to directly obtain the physical quantities.

[0021] <Step S4> In step S4, the visibility probability calculation unit 14 calculates the visibility probability for each target area or each target object based on the map information I1 and the line-of-sight information I2. At this time, the target area is divided into a grid, or each area is labeled and divided using a method such as semantic segmentation. The target object is a stationary obstacle or a moving obstacle included in the map information I1.

[0022] Here, using FIG. 4, a change in the recognition probability of the pedestrian P in the environment of FIG. 3, in which the vehicle V is traveling on a straight road and the pedestrian P is on the sidewalk to the left front, is explained. In FIG. 3(a), the distance from the vehicle V to the pedestrian P is long and the pedestrian P is not present in the driver's field of view F, so the probability that the driver recognizes the pedestrian P (visibility probability) is low, as shown in FIG. 4(a). In FIG. 3(b), the vehicle V approaches the pedestrian P and the pedestrian P is included in the field of view F, so the driver's pedestrian visibility probability increases sharply, as shown in FIG. 4(b). In FIG. 3(c), the driver is looking to the right front and the pedestrian P is not included in the field of view F, but unless the pedestrian P behaves erratically, the pedestrian P is in the vicinity of the position in FIG. 3(b) and the driver should recognize it as such, so the driver's pedestrian visibility probability is set to maintain a high value. However, in this case, the driver's pedestrian visibility probability is gradually decreased over time. In this way, the visibility probability calculation unit 14 calculates the visibility probability taking into consideration the time-series change in the line-of-sight information I2.

[0023] <Step S5> In step S5, the vehicle trajectory calculation unit 15 calculates a vehicle trajectory T along which the vehicle V is predicted to travel in the future based on the vehicle information I3 acquired in step S3. V Here, the vehicle trajectory T V is information including points representing the vehicle position and vehicle attitude (yaw angle) on a two-dimensional plane, and the running speed at each point. Note that instead of speed, the vehicle trajectory T V The vehicle trajectory T V There are known methods for calculating the vehicle trajectory T from geometry, from a state equation of a vehicle model, or from a machine learning model such as a neural network. V When calculating the vehicle trajectory T, the prediction accuracy may be improved by using the vehicle information I3 acquired in the past, rather than using only the vehicle information I3 at the current time. VFor example, when it is determined that a left turn is to be performed, the vehicle trajectory T is expressed by three areas: the lane in which the vehicle is traveling, the lane to which the vehicle is to turn, and the area connecting the vehicle lane and the lane to which the vehicle is to turn. V Let us assume that.

[0024] <Step S6> In step S6, the risk level calculation unit 16 calculates the risk level based on the map information I1 acquired in step S1, the visibility probability calculated in step S4, and the vehicle trajectory T V The probability (risk level) of dangerous driving occurring is calculated based on the above. Here, the risk level calculation process in this step will be specifically described with reference to Figs.

[0025] <<First example of dangerous driving>> 5 is a top view illustrating an example of a situation in which dangerous driving occurs. In this situation, the danger level calculation unit 16 first calculates a pedestrian trajectory T P Then, the vehicle trajectory T calculated in step S5 is predicted. V The predicted pedestrian trajectory T P When the intersection is between the intersections, the risk level calculation unit 16 sets the risk level to be high regardless of the driver's pedestrian visibility probability (see FIG. 4).

[0026] <<Second example of dangerous driving occurrence>> FIG. 6 is a top view illustrating another example of a situation in which dangerous driving occurs. In this situation, the danger level calculation unit 16 first detects a traffic light in the intersection ahead that is included in the map information I1. If there is a traffic light ahead, driving according to the signal instructions is required. Therefore, if the visibility probability of the traffic light calculated in step S4 is low, it can be estimated that the traffic light has not been confirmed. Therefore, the vehicle trajectory T estimated in step S5 is used as the vehicle trajectory T V However, if the vehicle crosses the stop line just before the intersection, the danger level calculation unit 16 sets the danger level to a high level.

[0027] <<Third example of dangerous driving occurrence>> FIG. 7 is a top view illustrating another example of a situation in which dangerous driving occurs. V The driver is planning to make a left turn along the road. In this case, the driver needs to check ahead, check for entrapment, and check the left turn destination in sequence. Therefore, the risk level calculation unit 16 first checks the driver's line of sight information I2 acquired in step S2, and judges whether the driver checks the area R on the left rear directly or through the mirror. Then, the risk level calculation unit 16 calculates the occurrence probability. For example, if the driver does not see the area R, the risk level calculation unit 16 judges that the occurrence probability of dangerous driving is high. A method may be used in which an arbitrary threshold is set and the calculated occurrence probability is judged to be high if it is equal to or higher than the threshold. If it is judged that the occurrence probability of dangerous driving is high, the risk level calculation unit 16 sets the risk level high. Needless to say, in countries and regions where vehicles are required to drive on the right side of the road, a similar judgment may be made by assuming a situation in which FIG. 7 is reversed horizontally.

[0028] In a situation similar to that shown in FIG. 7, the danger level calculation unit 16 calculates a safety area R S For example, even if it is determined based on the line-of-sight information I2 that the driver has not confirmed that an entrapment has occurred in the left rear area R in FIG. 7, it may be determined that the driver has confirmed that the area ahead is safe. In that case, the risk level calculation unit 16 may set a safety area R ahead of the vehicle V as shown in FIG. S If an obstacle exists in front of the vehicle V, a safety area R is set excluding the area where the obstacle exists. S In addition, lane information included in map information I1 obtained from the navigation system is used to set a safety area R including the area where the vehicle can travel. S may be set.

[0029] As shown in Fig. 8, the vehicle trajectory T V is the safe region R S If the pass through the safety area R S The probability of dangerous driving occurring is small within the safe area R SThe probability of occurrence of dangerous driving becomes high in the region beyond this range. Therefore, the danger level calculation unit 16 dynamically changes the probability of occurrence of dangerous driving.

[0030] For example, as shown in FIG. 9, when a vehicle V is in a safe area R S The probability of dangerous driving occurring is set to be small for the time when the vehicle is predicted to pass point P1 within the safety region R S At the time when it is predicted that the vehicle will pass points P2 and P3, the probability of dangerous driving occurring is gradually increased.

[0031] <Step S7> In step S7, the danger level calculation unit 16 judges whether the occurrence probability of dangerous driving (danger level) is equal to or higher than a threshold. If the occurrence probability is equal to or higher than the threshold, the process proceeds to step S8, and if it is lower than the threshold, the process ends. Note that in FIG. 2, in order to simplify the flowchart, the same threshold is used regardless of the type of obstacle, but the threshold for stationary obstacles and the threshold for moving obstacles may be separated and the latter may be set to a larger value.

[0032] <Step S8> In step S8, the output unit 17 warns the driver of the occurrence of dangerous driving via the HMI device 4. There are various methods for issuing the warning, such as issuing a warning sound or voice guidance using a speaker, or drawing on a liquid crystal panel or a HUD (Head Up Display), but the method is not important in the present invention.

[0033] <Modification of Example 1> In FIG. 2, when proceeding from step S7 to step S8, a uniform warning is issued regardless of the magnitude of the risky driving occurrence probability (risk level) calculated in step S6. However, multiple thresholds for distinguishing the risk level in step S7 may be prepared, and the degree of warning may be changed depending on the result of the distinction.

[0034] For example, a first threshold, a second threshold, and a third threshold may be prepared in ascending order, and dangerous driving occurrence probability (danger level) that is equal to or greater than the first threshold and less than the second threshold may be classified as a "near miss" occurrence, that is equal to or greater than the second threshold and less than the third threshold may be classified as a "near miss" occurrence, and that is equal to or greater than the third threshold may be classified as an "accident prediction" occurrence. A near miss occurrence is an event that is unlikely to lead to an accident such as a collision, but is likely to make the driver or pedestrian feel a sense of danger.

[0035] The danger level classified as "accident warning" includes the possibility of a serious accident, such as a collision between vehicle V and pedestrian P, and is dangerous driving that must be avoided without fail (see Figure 5). In this case, in order to reliably notify the driver of the "accident warning," for example, a loud speaker is used to provide an auditory warning, and a visual warning (see Figure 10) is also used, showing an image of a flashing "!" mark on the display at the position where a collision with pedestrian P is predicted, to prompt the driver to take the necessary action to avoid the accident, such as braking or steering.

[0036] The danger level classified as "near miss" is a case where sudden braking or steering occurs due to a delay in visually recognizing an obstacle or a traffic signal. In this case, a loud speaker sounds to notify the driver of the "near miss" audibly.

[0037] The danger level classified as "near miss sign" is driving that may actually be dangerous, even if it does not involve sudden braking or steering. This is the case where a collision check was not performed, as shown in Fig. 7. In this case, the speaker sounds at a low volume to notify the driver of the "near miss sign".

[0038] By utilizing the dangerous driving warning device of the present embodiment described above, it is possible to estimate the danger level based on time-series changes in the driver's gaze information, and to provide the driver with an appropriate warning according to the estimated danger level. EXAMPLES

[0039] Next, a dangerous driving warning device according to a second embodiment of the present invention will be described with reference to Fig. 11 to Fig. 14. Note that a duplicated description of points common to the first embodiment will be omitted.

[0040] As shown in FIG. 11, the dangerous driving warning device 1A of this embodiment has a cognitive grid map creation unit 18, which is a variant of the visibility probability calculation unit 14, in addition to the map acquisition unit 11, gaze information acquisition unit 12, vehicle information acquisition unit 13, vehicle trajectory calculation unit 15, danger level calculation unit 16, and output unit 17 described in the first embodiment.

[0041] Next, the warning generation process by the dangerous driving warning device 1A described above will be described with reference to the flowchart in Fig. 12. Note that since steps S11 to S13, S15, S17, and S18 in Fig. 12 are equivalent to steps S1 to S3, S5, S7, and S8 in Fig. 2, their description will be omitted, and steps S14 and S16, which are processes unique to this embodiment, will be described in detail.

[0042] In step S14, a cognitive grid map M1 is created based on the line-of-sight information I2 acquired in step S12. This cognitive grid map M1 is a map that projects the line-of-sight information I2 onto a grid-like two-dimensional plane with the vehicle center or fixed coordinates as the origin, and handles the driver's visual recognition area probabilistically. In addition, the cognitive grid map M1 is updated using the previously created cognitive grid map M1 and the line-of-sight information I2.

[0043] For example, Fig. 13 shows the time series changes in the perception grid map M1 with the center of the vehicle as the origin when turning left at an intersection, as in Fig. 7. Because the driver is looking ahead before entering the intersection, the driver can see the grid in front of the vehicle, so the visibility probability is high (white grid), as shown in Fig. 13(a). On the other hand, the driver cannot see the area behind the vehicle, so the visibility probability is low (black grid). The grids on the sides of the vehicle are areas that have been seen in the past, so the visibility probability is medium (gray grid), and the probability decreases over time.

[0044] Next, when the driver checks for an obstacle, the driver can see the sides of the vehicle through the mirror, and the probability of seeing the left rear and right rear becomes high (white grid) as shown in Figure 13(b). However, if the driver continues to check for an obstacle, the driver cannot see the front of the vehicle during that time, and the probability of seeing the front of the vehicle decreases over time (changing from white grid to gray grid) as shown in Figure 13(c).

[0045] In step S16, the risk level calculation unit 16 calculates the risk level based on the map information I1 acquired in step S11, the cognitive grid map M1 created in step S14, and the vehicle trajectory T calculated in step S15. V Based on this, the occurrence probability of dangerous driving is calculated, and the process proceeds to step S17.

[0046] In this step, the vehicle trajectory T is plotted on the cognitive grid map M1 (see FIG. 13). V (See Fig. 7) and project the vehicle trajectory T onto the grid with low visibility (gray grid or black grid). V If the sequence of points exists, it is determined that the occurrence probability (risk level) of dangerous driving is high. V If the obstacle's movement prediction point sequence comes into contact with the obstacle, the danger level is set to a high level.

[0047] In addition, in this step, the occupancy grid map M2 acquired from the map information I1 may be utilized and superimposed on the cognitive grid map M1 to calculate the risk of dangerous driving occurrence. For example, as shown in FIG. 14, the lower safety grid map M3 may be created by superimposing the upper cognitive grid map M1 (see FIG. 13) and the middle occupancy grid map M2. In the middle occupancy grid map M2, grids where no obstacles exist are colored white, and grids where obstacles exist are colored black. The cognitive grid map M1 may be created based on point cloud information acquired from an external recognition sensor, instead of being created from obstacle information prepared in advance.

[0048] By treating the grid probabilistically in this way, it is possible to easily calculate the probability of dangerous driving occurring, and at the same time, it is possible to easily classify "accident precursors," "near misses," and "near miss precursors" according to the probability of occurrence.

[0049] From the above, by creating a cognitive grid map M1 from the gaze information I2, the probability of dangerous driving occurring can be easily handled, and a warning according to the probability of dangerous driving can be output with high accuracy.

[0050] The above describes in detail an embodiment of the vehicle control device according to the present disclosure using the drawings, but the specific configuration is not limited to this embodiment, and even if there are design changes, etc., within the scope that does not deviate from the gist of the present disclosure, they are included in the present disclosure. [Explanation of symbols]

[0051] 100 Driving Assistance System 1. 1A Dangerous driving warning device 11 Map Acquisition Section 12 Gaze information acquisition unit 13 Vehicle information acquisition unit 14 Visibility Probability Calculation Unit 15 Vehicle trajectory calculation unit 16 Risk level calculation unit 17 Output section 18 Cognitive Grid Mapping Department 2. External recognition sensor 3 Vehicle Sensors 4 HMI device I1 Map Information I2 gaze information I3 vehicle information M1 cognitive grid map M2 Occupancy Grid Map M3 Safety Grid Map

Claims

1. a map acquisition unit that acquires map information including the location of an obstacle; A gaze information acquisition unit that acquires gaze information of a driver who drives a vehicle; a vehicle information acquisition unit that acquires vehicle information including a vehicle speed and a steering angle; a visibility probability calculation unit that calculates a visibility probability of the driver in an area visible to the driver by using the line-of-sight information; a vehicle trajectory calculation unit that calculates a trajectory of the vehicle using the vehicle information; a risk level calculation unit that calculates a risk level, which is a probability that dangerous driving will occur, using the map information, the visibility probability, and the vehicle trajectory; an output unit that outputs a warning to the driver when the danger level is equal to or greater than a given threshold; A dangerous driving warning device comprising:

2. 2. The dangerous driving warning device according to claim 1, 11. A dangerous driving warning device, comprising: a driver's line of sight information calculated based on the driver's line of sight;

3. 2. The dangerous driving warning device according to claim 1, A dangerous driving warning device characterized in that the visibility probability calculation unit calculates the visibility probability by probabilistically calculating that an obstacle included in the map information is included in the line of sight information.

4. 2. The dangerous driving warning device according to claim 1, The risk level calculation unit is Calculating an area in which the vehicle can travel from the map information; Calculating a safety area from the visibility probability and the drivable area; A dangerous driving warning device, characterized in that the danger level is calculated using the safety area and the vehicle trajectory.

5. 2. The dangerous driving warning device according to claim 1, The dangerous driving warning device is characterized in that the map acquisition unit acquires map information including obstacles and traffic rules.

6. 2. The dangerous driving warning device according to claim 1, The dangerous driving warning device is characterized in that the output unit outputs a warning whose content corresponds to the magnitude of the danger level.

7. 2. The dangerous driving warning device according to claim 1, The dangerous driving warning device is characterized in that the danger level calculation unit predicts a pedestrian's walking trajectory from pedestrian position information and speed information included in the map information, and calculates a danger level depending on whether the vehicle's trajectory intersects with the predicted pedestrian walking trajectory.

8. 2. The dangerous driving warning device according to claim 1, The danger level calculation unit estimates from the line of sight information whether or not the traffic rule information contained in the map information has been confirmed, and calculates a danger level using the estimation result and the vehicle's trajectory.

9. 2. The dangerous driving warning device according to claim 1, The danger level calculation unit, when the vehicle's trajectory is turning left or right, checks whether the vehicle is checking for an entrapment to the left or right rear based on the line of sight information, and calculates a danger level using the confirmation result.

10. 2. The dangerous driving warning device according to claim 1, The visibility probability calculation unit is a cognitive grid map creation unit that creates a cognitive grid map using the line of sight information, The dangerous driving warning device is characterized in that the danger level calculation unit calculates a danger level, which is a probability that dangerous driving will occur, using the map information, the cognitive grid map, and the vehicle trajectory.

11. 11. The dangerous driving warning device according to claim 10, The cognitive lattice map creation unit is characterized in that it probabilistically calculates the cognitive lattice map using time-series changes in the line of sight information.

12. 11. The dangerous driving warning device according to claim 10, The danger level calculation unit creates a safety grid map by superimposing an occupancy grid map and the cognitive grid map, and calculates the danger level based on the safety grid map.

13. An external recognition sensor that recognizes the surrounding environment of the vehicle; A vehicle sensor that acquires vehicle information; an HMI device that acquires a line of sight direction of a driver who drives the vehicle; A dangerous driving warning device that warns when the occurrence of dangerous driving is predicted; A vehicle equipped with The dangerous driving warning device includes: a map acquisition unit that acquires map information including the location of an obstacle; A line-of-sight information acquisition unit that acquires line-of-sight information of a driver who drives the vehicle; a vehicle information acquisition unit that acquires vehicle information including a vehicle speed and a steering angle; a visibility probability calculation unit that calculates a visibility probability of the driver in an area visible to the driver by using the line-of-sight information; a vehicle trajectory calculation unit that calculates a trajectory of the vehicle using the vehicle information; a risk level calculation unit that calculates a risk level, which is a probability that dangerous driving will occur, using the map information, the visibility probability, and the vehicle trajectory; an output unit that outputs a warning to the driver when the danger level is equal to or greater than a given threshold; A vehicle comprising:

14. a map acquisition step of acquiring map information including the location of an obstacle; A gaze information acquiring step of acquiring gaze information of a driver who drives a vehicle; A vehicle information acquisition step of acquiring vehicle information including a vehicle speed and a steering angle; a visibility probability calculation step of calculating a visibility probability of the driver in an area visible to the driver by using the line-of-sight information; a vehicle trajectory calculation step of calculating a trajectory of the vehicle using the vehicle information; a risk level calculation step of calculating a risk level, which is a probability that dangerous driving will occur, using the map information, the visibility probability, and the vehicle trajectory; a warning step of warning the driver if the danger level is equal to or greater than a given threshold; A dangerous driving warning method comprising:

Images