JP5577609B2 - Driving assistance device - Google Patents

Description

  The present invention relates to a driving support device and a driving support method.

  Conventionally, when a driver's gaze direction deviates from a predetermined gaze direction region, a technique for estimating a driver's action intention based on the deviation of the driver's gaze direction is known (Patent Document 1). .

JP 2002-331849 A

  However, in the prior art, when the driver's line-of-sight direction moves greatly until it deviates from the predetermined line-of-sight direction area, and the state where the driver's line-of-sight direction deviates from the predetermined line-of-sight direction area continues for a predetermined time or longer, Since it is intended to estimate the driver's behavior intention, it takes time to estimate the driver's behavior intention, and in some cases, driving support cannot be appropriately performed.

  The problem to be solved by the present invention is to provide a driving support device capable of appropriately estimating the driver's behavior intention.

The present invention predicts a plurality of behaviors that the vehicle can take by the driver's operation according to the situation around the vehicle , repeatedly calculates the center of gravity of the driver's gaze direction distribution based on the driver's gaze direction, The difference in the vehicle width direction between the calculated center of gravity distribution of the driver's line of sight and a predetermined initial center of gravity is repeatedly calculated as a deviation amount in the vehicle width direction, and the deviation amount in the vehicle width direction is predetermined. By determining whether or not a threshold value corresponding to the behavior of the vehicle is exceeded, the behavior selected by the driver from the plurality of predicted behaviors is estimated, thereby solving the above-described problem.

  According to the present invention, in order to appropriately estimate the driver's action intention, it is possible to appropriately support the driver's driving.

It is a block diagram which shows the structure of the vehicle carrying the driving assistance system which concerns on 1st Embodiment. It is a figure for demonstrating the mutual relationship of each structure which builds the driving assistance system which concerns on 1st Embodiment. It is a flowchart which shows the driving assistance process in the driving assistance device which concerns on 1st Embodiment. It is a figure which shows the example of a scene where the vehicle is approaching the preceding vehicle and the vehicle is traveling on the second traffic zone (central traveling lane) of the road of three lanes on one side. It is a figure which shows an example of distribution of a driver | operator's gaze direction. It is a figure which shows the shift | offset | difference of the distribution of a gaze direction in the example of a scene where a vehicle changes to a right lane after a straight run. It is a flowchart which shows the behavior candidate prediction process which concerns on 1st Embodiment. It is a figure for demonstrating the estimation method of the approach state with a preceding vehicle. It is a figure which shows an example of the table which recorded the scene information which concerns on 1st Embodiment. It is a flowchart which shows the behavior estimation process which concerns on 1st Embodiment. It is a figure which shows an example of the time series change of the deviation | shift amount _Xd of the horizontal direction (X-axis direction) in the example of a scene where a driver | operator changes the vehicle 1 to a right lane. It is the schematic of the route guidance information displayed on the display with which a navigation apparatus is provided in the example of a scene of a route guidance left turn instruction | indication (no side road). It is the schematic of the route guidance information displayed on the display with which a navigation apparatus is provided in the example of a scene of a route guidance left turn instruction | indication (with a side road). It is a flowchart which shows the driving assistance process in the driving assistance device which concerns on 2nd Embodiment. It is a flowchart which shows the behavior estimation process which concerns on 2nd Embodiment. It is a flowchart which shows the behavior ratio calculation process which concerns on 2nd Embodiment. It is a figure which shows an example of the table which recorded the scene information which concerns on 2nd Embodiment.

<< First Embodiment >>
DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, a first embodiment of the invention will be described with reference to the drawings.
In the first embodiment, a vehicle having a driving support system that supports driving by a driver will be described as an example. Here, FIG. 1 is a block diagram showing a configuration of the vehicle 1 equipped with the driving support system according to the first embodiment. As shown in FIG. 1, the vehicle 1 includes a driving support device 10, an infrared illumination 20, a camera 30, a vehicle controller 40, and a navigation device 50.

  FIG. 2 is a diagram for explaining the interrelationships of the components that construct the driving support system according to the first embodiment. As shown in FIG. 2, the near-infrared illumination 20 irradiates the driver's head (including the face) from the front, and the camera 30 images the front of the driver's head irradiated by the near-infrared illumination 20. . In this way, by irradiating the front of the driver's head with the near-infrared illumination 20, the camera 30 can appropriately capture the front of the driver's head even in a dark place such as at night or in a tunnel. it can. Image information captured by the camera 30 is transmitted to the driving support device 10.

  In addition to the image information transmitted from the camera 30, the driving support device 10 acquires vehicle information and surrounding information from the vehicle controller 40, and acquires surrounding information from the navigation device 50. The vehicle information includes various information such as the vehicle speed, the accelerator opening, the brake signal, and the steering angle of the steering, and the surrounding information includes various information such as the inter-vehicle distance to the preceding vehicle, the current position of the vehicle 1, and the traveling direction. Contains information. Then, the driving support device 10 estimates the behavior of the vehicle 1 selected by the driver based on these pieces of information, and transmits the estimated behavior information to the vehicle controller 40. The vehicle controller 40 controls various devices so as to perform driving support according to the behavior information. Thus, each structure which constructs the driving support system concerning this embodiment is related mutually as mentioned above, and can provide suitable driving support based on the estimated behavior information. Each of these components is connected by a CAN (Controller Area Network) or other vehicle-mounted LAN, and can exchange information with each other.

  Hereinafter, each component of the vehicle 1 will be described in detail.

  As shown in FIG. 2, the infrared illumination 20 and the camera 30 are installed at a position where an image of the front of the driver's head can be taken, such as in the front of the driver, in the lower meter panel, or near the steering column (not shown). . Note that the infrared illumination 20 and the camera 30 may be slightly shifted left and right as long as the front of the driver's head can be imaged. The camera 30 includes a CCD (Charge Coupled Device) having a sensitivity distribution from visible light to near-infrared light, a CMOS (Complementary Metal Oxide Semiconductor), and other image sensors as an image sensor, and is located in front of the driver's head. A lens having an angle of view so that the image can be captured. Note that a filter may be installed in the camera 30 in order to minimize the influence of sunlight.

  The vehicle controller 40 acquires various information related to the vehicle 1. Specifically, the vehicle controller 40 includes vehicle speed information from the vehicle speed sensor, an inter-vehicle distance from a front sensor that detects a front area of the vehicle 1 to a preceding vehicle, an accelerator opening from the accelerator sensor, and a brake signal from the brake sensor. And various information such as a steering angle by a steering operation from the steering angle sensor. Then, the vehicle controller 40 transmits the acquired information to the driving support device 10. Moreover, the vehicle controller 40 acquires the behavior information of the vehicle 1 estimated to be selected by the driver from the driving support device 10 and controls various devices based on the behavior information, thereby supporting the driving of the driver. .

  The navigation device 50 includes a GPS (Global Positioning System) module and a communication device, and acquires various types of information such as the current position and traveling direction of the vehicle 1. The navigation device 50 transmits the acquired information to the driving support device 10. The navigation device 50 has a route guidance function for guiding the travel route of the vehicle 1 based on map information and road information stored in the memory of the navigation device 50 and information acquired by a GPS module or the like.

  The driving support device 10 includes a ROM (Read Only Memory) that stores a program for estimating the behavior of the vehicle 1 selected by the driver, a CPU (Central Processing Unit) that executes the program stored in the ROM, It is composed of a RAM (Random Access Memory) that functions as an accessible storage device.

  The driving support device 10 implements each function of a gaze direction detection function, a behavior candidate prediction function, a behavior estimation function, a driver identification function, and a driving history learning function by executing a program stored in the ROM by the CPU. Below, each function with which the driving assistance device 10 is provided is demonstrated.

  The gaze direction detection function of the driving support device 10 detects the gaze direction of the driver from the image information transmitted from the camera 30. The method for detecting the gaze direction by the gaze direction detection function is not particularly limited. For example, the gaze direction detection function acquires image information in front of the driver's head imaged in time series by the camera 30, performs image processing on the acquired image information, and detects a plurality of characteristic parts of the driver's face. . The line-of-sight direction detection function estimates the driver's face direction based on the positional relationship between the plurality of detected characteristic parts, cuts out the driver's eye area from the image information, and performs image processing. Even the eye feature point including the pupil position is detected. Further, the gaze direction detection function estimates the gaze direction of the driver with reference to the camera 30 based on the estimated driver's face orientation and the detected eye feature point. And the gaze direction detection function estimates the logical front assumed that the driver recognizes the front based on the estimated time-series change of the gaze direction of the driver, and uses the estimated logical front, Correct the driver's gaze direction. Thereby, even when the installation position of the camera 30 is not in front of the driver, the gaze direction detection function can automatically acquire the front direction for each driver and appropriately detect the driver's gaze direction. it can. Note that the method for detecting the driver's line-of-sight direction is not limited to the above example. For example, the driver's line-of-sight direction may be detected using a plurality of stereo cameras or the like having a distance measuring function.

  The gaze direction detection function detects such a driver's gaze direction as, for example, coordinate data centered on the logical front of the driver. Then, the gaze direction detection function stores the detected coordinate data of the gaze direction in the RAM of the driving support device 10. Note that the RAM of the driving support apparatus 10 holds (stores) coordinate data in the line-of-sight direction for a certain time in the order of detection.

  The behavior candidate prediction function of the driving support device 10 estimates the vehicle state based on the vehicle information and surrounding information acquired from the vehicle controller 40 and the surrounding information acquired from the navigation device 50, and the vehicle controller 40 and the navigation device 50. Based on the surrounding information acquired from the above, the surrounding situation is estimated. Here, examples of the vehicle state include an approaching state with a preceding vehicle, a straight traveling state, an acceleration state, and a deceleration state. Further, examples of the surrounding situation include a road situation where the vehicle 1 is traveling, a situation of presence / absence of a side road around the vehicle 1, and the like. The behavior candidate prediction function predicts a plurality of behaviors that can be taken by the vehicle 1 by a driver's operation in the estimated vehicle state and surrounding situation.

  The behavior estimation function of the driving support device 10 calculates the centroid coordinates of the distribution of the gaze direction coordinate data at a predetermined time as the initial centroid based on the gaze direction coordinate data for a certain time detected by the gaze direction detection function. The difference between the calculated initial center of gravity and the center of gravity coordinates of the distribution of the coordinate data in the line-of-sight direction after the initial center of gravity is calculated is obtained as a shift amount. Then, the behavior estimation function estimates the behavior selected by the driver from the plurality of behaviors predicted by the behavior candidate prediction function based on the deviation amount.

  The driver identification function of the driving assistance device 10 identifies the driver based on the driver image information transmitted from the camera 30. Note that by using the driver information of the driver identified by the driver identification function, information based on the estimated behavior of the vehicle 1 can be presented to the driver at the timing, frequency, or method set by the driver.

  The driving history learning function of the driving support device 10 determines the behavior of the vehicle 1 selected by the driver in a predetermined vehicle state and surrounding situation. The driving history learning function stores the behavior of the vehicle 1 selected by the driver as the driving history for each vehicle state and surrounding situation and for each driver. Further, as will be described later, the driving history learning function calculates a behavior ratio of the vehicle 1 selected by the driver in a predetermined vehicle state and surrounding situation as a behavior ratio based on the stored driving history.

  Next, driving support processing in the driving support device 10 according to the first embodiment will be described with reference to the flowchart shown in FIG. FIG. 3 is a flowchart showing a driving support process in the driving support device 10 according to the first embodiment. FIG. 4 is a diagram illustrating an example of a scene in which the vehicle 1 is approaching the preceding vehicle and the vehicle 1 is traveling on the second traffic zone (central traveling lane) of a three-lane road on one side. In the following, as shown in FIG. 4, an example of a scene is assumed in which the vehicle 1 is approaching a preceding vehicle and the vehicle 1 is traveling on the second lane (central lane) of a three-lane road. To explain. In addition, the driving assistance process demonstrated below is performed by the driving assistance apparatus 10. FIG.

  First, in step S1, image information obtained by imaging the front of the driver's head transmitted from the camera 30 is acquired by the gaze direction detection function of the driving support device 10, and based on the acquired image information, the driver's head information is acquired. Coordinate data in the line-of-sight direction is detected. The detected coordinate data is stored in a RAM provided in the driving support device 10.

  Here, FIG. 5 is a diagram illustrating an example of the distribution of the gaze direction detected by the gaze direction detection function, and FIG. 6 is a diagram of the driver in an example of a scene in which the vehicle 1 changes to the right lane after traveling straight ahead. It is a figure which shows the shift of the distribution of a gaze direction. When the vehicle 1 travels straight ahead, as shown in FIG. 5, the driver's line-of-sight direction is distributed in the vicinity of the logical front assumed to be recognized by the driver as the front. In addition, as shown in FIG. 6, when the driver tries to change the vehicle 1 to the right lane after the vehicle 1 approaches the preceding vehicle after traveling straight ahead, the driver's line-of-sight direction is straight traveling. It is distributed in the right direction, which is the lane change direction, rather than the sight line direction at the time. The gaze direction detection function detects the gaze directions of these drivers as coordinate data in time series and stores them in the RAM provided in the driving support device 10.

  Next, in step S <b> 2, a plurality of behaviors that the vehicle 1 can take are predicted by the behavior candidate prediction function of the driving support device 10. Specifically, the vehicle state of the vehicle 1 is estimated based on the vehicle information and surrounding information acquired from the vehicle controller 40 and the surrounding information acquired from the navigation device 50, and is acquired from the vehicle controller 40 or the navigation device 50. The surrounding situation is estimated based on the surrounding information. Furthermore, in the estimated vehicle state and surrounding conditions, a plurality of behaviors that the vehicle 1 can take are predicted by the operation of the driver. Here, FIG. 7 is a flowchart showing the behavior candidate prediction process in step S2. The contents of the behavior candidate prediction process in step S2 will be described below along the flowchart of FIG.

  First, in step S <b> 201, the vehicle state of the vehicle 1 is estimated based on the vehicle information and surrounding information acquired from the vehicle controller 40 and the surrounding information acquired from the navigation device 50. Here, FIG. 8 is a diagram for explaining a method of estimating the approach state with the preceding vehicle. For example, an approaching state with a preceding vehicle, which is an example of a vehicle state, can be estimated based on changes in time-series vehicle speed information and inter-vehicle distance, as shown in FIG. Specifically, as shown in FIG. 8, the vehicle speed of the vehicle 1 increases as time passes, or the change (increase / decrease) in the vehicle speed with time passes is small, and When the inter-vehicle distance between the vehicle 1 and the preceding vehicle decreases with time, the vehicle 1 is estimated to be in a state of approaching the preceding vehicle. The vehicle state estimated by the behavior candidate prediction function is not limited to the approaching state with the preceding vehicle, and the approaching state estimation method with the preceding vehicle is not limited to the above method. For example, it is a communication apparatus which the navigation apparatus 50 has, Comprising: The approach state with a preceding vehicle can be estimated by using the communication apparatus which communicates between other vehicles.

  Next, in step S202, the surrounding situation of the vehicle 1 is estimated based on the surrounding information acquired from the vehicle controller 40 or the navigation device 50. In the example of the scene illustrated in FIG. 4, for example, based on road information transmitted from the navigation device 50, the road on which the vehicle 1 is traveling is estimated to be a three-lane road on one side. In addition, for example, by acquiring information from a communication device provided in the navigation device 50 that communicates with a roadside machine or the like, it is estimated that the vehicle 1 is traveling in the second traffic zone (central traveling lane) of the three lanes on one side. The Steps S201 and 202 may be performed in parallel, or step S202 may be performed first. The method for estimating the surrounding situation by the behavior candidate prediction function is not particularly limited. For example, the lane edge of the lane in which the vehicle 1 travels is detected using a camera installed on the upper part of the front window of the vehicle 1. It may be estimated which road of a road having a plurality of lanes is traveling.

  In step S203, the behavior candidate prediction function predicts a plurality of behaviors that can be taken by the vehicle 1 by the driver's operation in the estimated vehicle state and surrounding situation. The prediction of the plurality of behaviors is performed based on scene information based on the plurality of behaviors that can be taken by the vehicle 1 by the operation of the driver in the estimated vehicle state and surrounding situation.

  Here, FIG. 9 is a diagram illustrating an example of a table in which scene information according to the first embodiment is recorded. In the table shown in FIG. 9, a value corresponding to the possibility of the driver selecting the leftward movement and the possibility of the rightward movement being selected by the driver for each vehicle state / ambient situation. A value corresponding to is set. For example, as in the example of the scene shown in FIG. 4, the vehicle 1 is in a state of approaching the preceding vehicle, and the vehicle 1 travels on the second lane of the road with three lanes on one side (central lane). When the vehicle 1 is in operation, the vehicle 1 is changed to the left lane, the lane is changed to the right lane, or the preceding vehicle that is neither of the following is followed by the driver's operation. Is expected to be adopted. Further, in the example of the scene shown in FIG. 4, the vehicle 1 is traveling on the second traffic zone (central traveling lane) of the road with three lanes on one side, so that the driver may change the vehicle 1 to the left lane. The possibility that the driver will change the vehicle 1 to the right lane is predicted to be about the same. Therefore, in the table shown in FIG. 9, in the surrounding situation where the vehicle 1 is in the vehicle state of “approaching the preceding vehicle” and the vehicle 1 is traveling in “one lane on the one side, second lane”, the driver Is set to a value “0.5” corresponding to the possibility that the leftward movement will be selected and a value “0.5” corresponding to the possibility that the rightward movement will be selected by the driver. ing.

  In this way, in the table shown in FIG. 9, values corresponding to the possibility that each behavior is selected are set for each vehicle state and surrounding situation, so the behavior candidate prediction function is the table shown in FIG. 9. By referring to the scene information set to, it is possible to predict a plurality of behaviors that can be taken by the vehicle 1 by the driver's operation in the estimated vehicle state and surrounding conditions. For example, the value “0.5” corresponding to the possibility of the leftward movement being selected by the driver in the vehicle state “approaching preceding vehicle” and the surrounding situation “three lanes on one side, second lane”, and the driver By acquiring scene information with a value of “0.5” corresponding to the possibility that the rightward movement is selected by the vehicle, the vehicle 1 changes the lane to the left lane or changes to the right lane by the driver's operation. It can be predicted that any of the three behaviors of changing lanes or following the preceding vehicle can be taken.

  In addition, when the vehicle 1 is in a state of approaching the preceding vehicle and the vehicle 1 is traveling on the first lane (left lane) of the road of two lanes on one side or one lane on one side, It is not assumed that the driver moves the vehicle 1 in the left direction (changes the lane to the left lane). Therefore, in the vehicle state “approaching preceding vehicle” in the table shown in FIG. 9 and the surrounding condition “two lanes on one side, first lane” or “one lane on one side, first lane”, the driver moves leftward. The value corresponding to the possibility that the movement is selected is set to “0”. Thereby, the behavior candidate predicting function allows the driver to move leftward by the driver in the vehicle state “approaching preceding vehicle” and the surrounding situation “two lanes on one side, first lane” or “one lane on one side, first lane”. By acquiring scene information of the value “0” corresponding to the possibility of selecting movement and the value “0.5” corresponding to the possibility of selecting movement in the right direction by the driver, It is predicted that the vehicle 1 can take one of two behaviors by changing the lane to the right lane or following the preceding vehicle by an operation of the person.

  In addition, the value according to the possibility of selecting the movement of the driver in the left-right direction acquired as the scene information is used when calculating a threshold value as a determination criterion for estimating the behavior of the vehicle 1 as will be described later. Used as a weight. Moreover, the table shown in FIG. 9 is an example, for example, it is good also as a table structure which can determine the behavior which the vehicle 1 can take by driver | operator's operation from one of a vehicle state or a surrounding condition.

  In step S203, after predicting a plurality of behaviors that the vehicle 1 can take with the behavior candidate prediction function, the behavior candidate prediction processing in step S2 is terminated. Note that if a plurality of behaviors that can be taken by the vehicle 1 due to the driver's operation cannot be predicted from the estimated vehicle state and surrounding conditions, the driving support processing in the driving support device 10 of FIG. 3 is terminated.

  Next, the behavior estimation process in step S3 will be described. In step S3, the behavior estimation function of the driving support device 10 estimates the behavior selected by the driver from the plurality of behaviors predicted to be taken by the vehicle 1 by the driver's operation. Here, FIG. 10 is a flowchart showing the behavior estimation processing in step S3. The contents of the behavior estimation process in step S3 will be described below along the flowchart of FIG.

First, in step S301, the barycentric coordinates (X _c , Y _c ) of the distribution of the coordinate data (X _i , Y _i ) in the gaze direction for a certain time detected by the gaze direction detection function are calculated.

ここで、式（１）および式（２）における“ｎ”は一定時間分の視線方向の座標データ（Ｘ_ｉ，Ｙ_ｉ）の数を示している。 Here, a method of calculating the barycentric coordinates (X _c , Y _c ) by the behavior estimation function will be described. As shown in FIG. 5, the coordinate data of the line-of-sight direction detected by the line-of-sight direction detection function is expressed in a coordinate system centered on the logical front assumed to be recognized by the driver as the front. Further, as described above, the coordinate data of the line-of-sight direction is stored in the RAM provided in the driving support device 10 for a certain time in the order of detection. The behavior estimation function is a distribution of the gaze direction coordinate data (X _i , Y _i ) from the line of gaze direction coordinate data (X _i , Y _i ) for a certain time stored in the RAM provided in the driving support device 10. obtained based the barycentric coordinates _(X c, _{Y c)} of the following formula (1) and the following formula (2). The behavior estimation function calculates the barycentric coordinates (X _c , Y _c ) of the distribution of the coordinate data (X _i , Y _i ) in the line-of-sight direction at a predetermined cycle.

Here, “n” in Expression (1) and Expression (2) indicates the number of coordinate data (X _i , Y _i ) in the line-of-sight direction for a certain time.

Next, in step S302, based on the barycentric coordinates (X _c , Y _c ) calculated in step S301, a shift amount that is the magnitude of the shift in the driver's line-of-sight direction is calculated. Here, the barycentric coordinates (X _c , Y _c ) are calculated at a predetermined period before a plurality of behaviors that the vehicle 1 can take are predicted by the behavior candidate prediction function. In step S302, the behavior estimation function uses the center of gravity coordinates (X _c , Y _c ) calculated at the time when a plurality of behaviors that can be taken by the vehicle 1 are predicted by the behavior candidate prediction function as the initial center of gravity (X ₀ , Y ₀ ). Asking. Moreover, behavior estimation function, polarized the difference between the initial centroids _(X 0, _{Y 0)} the center coordinates _(X c, _{Y c)} of the calculated line-of-sight direction after the calculation of the initial center of gravity _(X 0, _{Y 0)} Obtained as transfer amounts X _d and Y _d .

Here, a method of calculating the deviation amount by the behavior estimation function will be described. The behavior estimation function calculates the initial coordinates of the gaze direction coordinate data (X _i , Y _i ) stored in the RAM of the driving support device 10 after calculating the initial center of gravity (X ₀ , Y ₀ ) and the initial coordinates. Differences from the center of gravity (X ₀ , Y ₀ ) are calculated as deviation amounts X _d , Y _d . Specifically, the behavior estimation function calculates the shift amounts X _d and Y _d based on the following formula (3) and the following formula (4).

For example, in the example of the scene shown in FIG. 4, when the driver tries to change the vehicle 1 to the right lane, the driver confirms safety in the right direction before actually changing the vehicle 1 to the right lane. To do. Therefore, as shown in FIG. 6, if the center of gravity coordinates of the distribution in the line of sight of the driver during straight traveling is the initial center of gravity (X ₀ , Y ₀ ), the driver 1 before the vehicle 1 changes lanes to the right lane The line-of-sight direction is distributed to the right of the initial center of gravity (X ₀ , Y ₀ ). That is, the center-of-gravity coordinates of the distribution in the line-of-sight direction before the vehicle 1 changes to the right lane shifts to the right of the initial center of gravity (X ₀ , Y ₀ ). Therefore, as shown in FIG. 6, the behavior estimation function uses the difference between the centroid coordinates of the distribution in the gaze direction after the calculation of the initial centroid and the initial centroid (X ₀ , Y ₀ ) as the shift amounts X _d , Y _d. calculate. In the present embodiment, shift amount _{X d} in the lateral direction (X axis direction) of the shift amount _X d, _{Y d,} the distribution of the line of sight direction after the calculation of the initial centroids _(X 0, _{Y 0)} the barycentric coordinates the initial centroids (X _{0, Y 0)} when the shift to the right of takes a positive value, the initial initial centroids (X _{0, Y 0)} barycentric coordinates of the line of sight direction of the distribution after calculation of When shifting to the left of the center of gravity (X ₀ , Y ₀ ), a negative value is assumed. As described above, when the shift amount _Xd in the lateral direction (X-axis direction) is used, when the shift amount _Xd is close to 0 with little change in the driver's line-of-sight direction, the behavior estimation function is Can be estimated to go straight. On the other hand, when the driver's line-of-sight direction shifts to the right and the shift amount _Xd increases in the positive direction, the behavior estimation function indicates that the driver changes the lane of the vehicle 1 to the right or turns right. Can be estimated. Further, when the driver's line-of-sight direction shifts to the left and the shift amount _Xd decreases in the negative direction, the behavior estimation function estimates that the driver changes the lane of the vehicle 1 to the left or turns left. it can.

In step S303, the behavior estimation function, the calculated shift amount _{X d} and the threshold value _t r, is compared with _{t l,} shift amount _{X d} is determined whether exceeds a predetermined threshold value _t r, _{t l} Thus, the behavior of the vehicle 1 selected by the driver is estimated. Here, in the present embodiment, the threshold value _tr is a positive value used to estimate the movement of the vehicle 1 in the right direction, and the threshold value t ₁ is used to estimate the movement of the vehicle 1 in the left direction. It is assumed that it is a negative value used for. Specifically, the behavior estimation function, shift amount _X d, and shift amount in the lateral direction (X axis direction) of the _{Y d X d} and the threshold value _t r, and _{t l} compared respectively, laterally ( when shift amount X _d in the X-axis direction) is larger than the threshold value t _r corresponding to the right direction, the vehicle 1 is assumed to move in the right direction, the offset of X _d corresponds to the left is smaller than the threshold t _l, the vehicle 1 is assumed to move in the left direction. Moreover, behavior estimation function, shift amount X _d Which threshold t _r, if t _l not be exceeded, the vehicle 1 is estimated to straight.

変化幅ｓは偏移量Ｘ_ｄ，Ｙ_ｄが採りうる変化幅であり、実験などにより決定される任意の値である。また、重みｗ_ｒ，ｗ_ｌは、推定した車両状態および周囲状況において運転者により選択される車両１の挙動に対応しており、推定した車両状態および周囲状況おいて運転者に選択される可能性に応じて設定される値である。すなわち、右方向に対応する重みｗ_ｒは、推定した車両状態および周囲状況において、運転者が右方向への移動を選択する可能性に応じて設定され、左方向に対応する重みｗ_ｌは、推定した車両状態および周囲状況において、運転者が左方向への移動を選択する可能性に応じて設定される。なお、重みｗ_ｒ，ｗ_ｌは、図９に示すテーブルに設定されているため、挙動推定機能は、ステップＳ２において取得した場面情報を参照して、重みｗ_ｒ，ｗ_ｌを取得することができる。 Here, the driver selects threshold t _r as a criterion in estimating the behavior of the vehicle 1, the method for calculating the t _l will be described. Behavior estimation function is calculated based on the threshold value t _r corresponding to the right direction in formula (5), is calculated on the basis of the threshold value t _l corresponding to the left direction in the following equation (6).

The change width s is a change width that the deviation amounts X _d and Y _d can take, and is an arbitrary value determined by experiments or the like. Also, the weights w _r and w _l correspond to the behavior of the vehicle 1 selected by the driver in the estimated vehicle state and surrounding situation, and can be selected by the driver in the estimated vehicle state and surrounding situation. This value is set according to the sex. That is, the weights w _r corresponding to the right direction, in the estimated vehicle condition and surrounding condition, is set in accordance with the possibility that the driver selects the rightward movement, the weight w _l corresponding to the left direction, It is set according to the possibility that the driver will select the leftward movement in the estimated vehicle state and surrounding conditions. Since the weights w _r and w _l are set in the table shown in FIG. 9, the behavior estimation function can acquire the weights w _r and w _l with reference to the scene information acquired in step S2. it can.

The formula (5), as shown in (6), and the threshold t _l corresponding to the threshold t _r and the left direction corresponding to the right direction, the weights w _r, to be considered w _l, the same in absolute value because it is set to a value, the weights w _r in accordance with the _{selectability,} by calculating the threshold value t _r, t _l with w _l, estimates the behavior of the vehicle 1 in accordance with the selectability can do. As a result, it is possible to reduce the fact that a behavior that is unlikely to be selected in the estimated vehicle state and surrounding situation is erroneously selected by the driver.

Here, FIG. 11 is a diagram illustrating an example of a time-series change of the shift amount _Xd in the lateral direction (X-axis direction) in an example of a scene where the driver changes the vehicle 1 to the right lane. As illustrated in FIG. 11, the behavior estimation function calculates the shift amount X _d at a predetermined period from time T1 when the initial center of gravity (X ₀ , Y ₀ ) is calculated. In the scene example shown in FIG. 11, the driver tries to change the vehicle 1 to the right lane, and the deviation amount _Xd increases in the plus direction (right direction) as time elapses. Then, at time T2, the offset of X _d is detected that exceeds the threshold value t _r corresponding to the right direction. Thereby, the behavior estimation function estimates that the behavior of the vehicle 1 selected by the driver is a lane change to the right lane.

As described above, when estimating the behavior of the vehicle 1 selected by the driver using the behavior estimation function, the shift amount _Xd is used, so that the driving can be performed even at an early stage of the driver's selection behavior with little movement in the line of sight. The behavior of the vehicle 1 selected by the person can be estimated. For example, conventionally, the time when the driver's line-of-sight direction deviates from a predetermined line-of-sight direction region and the deviation state continues for a predetermined time or longer (the change (absolute value) of the shift amount _Xd in the present invention is large). However, according to the present embodiment, as shown in FIG. 11, the change (absolute value) in the deviation amount _Xd cannot be estimated. At a time T2 when is relatively small, the behavior of the vehicle 1 by the operation of the driver can be estimated.

  Even if the behavior estimation function cannot estimate the behavior of the vehicle 1 selected by the driver as early as time T2, as shown in FIG. 11, the driver usually changes the lane. In order to confirm the direction in which the lane change is performed immediately before, the behavior estimation function is performed before the time T4 when the vehicle 1 changes the lane to the right lane by the driver's operation (for example, about 0.5 seconds before to about 1 second before) At time T3, the behavior of the vehicle 1 selected by the driver can be estimated.

  Further, a behavior that is not predicted to be taken by the vehicle 1 by the behavior candidate prediction function, that is, a behavior in which a value corresponding to a possible application selected in the scene information is set to 0 is not determined in step S303.

  In step S303, after estimating the behavior of the vehicle 1 selected by the driver by the behavior estimation function, the driving support device 10 transmits the estimated behavior information of the vehicle 1 to the vehicle controller 40, and the driving in the driving support device 10 is performed. The support process ends.

  By using the behavior information of the vehicle 1 estimated in this way, the driving support device 10 controls the vehicle 1 to perform a preparatory operation for the driver's operation via the vehicle controller 40 or the driver. It is possible to assist the driver's driving by presenting information to the driver at an early stage. For example, when it is predicted from the estimated behavior of the vehicle 1 that the vehicle 1 will change to the right lane, the vehicle controller 40 that has acquired the behavior information from the driving support device 10 has the right side of the vehicle 1 as shown in FIG. By controlling a detection device that detects obstacles around the vehicle so as to detect other vehicles approaching from behind, or by automatically displaying the driver's blind spot with a monitor to assist the driver's blind spot, The driver can be made aware of the presence of another vehicle approaching from the right rear.

  In addition, for example, in a scene where the driver's brake operation is predicted from the estimated behavior of the vehicle 1, the vehicle controller 40 that acquires the behavior information from the driving support device 10 determines the brake hydraulic pressure in advance of the driver's brake operation. By raising the position, it is possible to support the driver's brake operation. In a scene where the driver's steering operation is predicted from the estimated behavior of the vehicle 1, the vehicle controller 40 that acquires the behavior information from the driving support device 10 turns the suspension of the vehicle 1 in advance of the driver's steering operation. By changing according to the direction, the driver's steering operation can be supported.

  Next, as shown in FIG. 12, the driving support processing in the driving support device 10 according to the first embodiment is performed before or after the original route in a situation where the route guidance information of the left turn instruction is presented from the vigation device 50. In the following description, it is assumed that there is no route (side road) in the left turn direction that may be erroneously selected by the driver. FIG. 12 is a schematic diagram of route guidance information displayed on a display provided in the navigation device 50 in an example of a route guidance left turn instruction (no side street).

  First, in step S1, the driver's gaze direction coordinate data is detected by the gaze direction detection function as in the above-described scene example. The detected coordinate data is stored in a RAM provided in the driving support device 10.

  Next, in the behavior candidate prediction process in step S2, a plurality of behaviors that the vehicle 1 can take are predicted by the behavior candidate prediction function. First, in step S201, it is estimated that the vehicle 1 is in a straight traveling state based on vehicle information such as a vehicle speed and a steering angle acquired from the vehicle controller 40.

  Next, in step S202, the route guidance information of the left turn instruction “turn left on the road ahead” transmitted from the navigation device 50 is acquired by the behavior candidate prediction function. When the route guidance information for the left turn instruction is acquired, the behavior candidate prediction function is based on the vehicle position information, the map information, and the road information acquired from the navigation device 50, and the driver is positioned before or behind the original guidance route. It is estimated whether there is a route (side road) that may be selected by mistake. In the example of the scene shown in FIG. 12, there is no route (side road) that may be erroneously selected by the driver before or behind the original guidance route, so the behavior candidate prediction function includes the vehicle position information, Based on the map information and the road information, it is estimated that there is no route (side road) that may be erroneously selected by the driver before or behind the original guidance route. When estimating whether there is a route (side road) that may be erroneously selected by the driver before or behind the original guidance route using the behavior candidate prediction function, an appropriately set threshold value or the like is used. For example, there is a path (side road) that may be erroneously selected by the driver by determining whether there is another side road within a predetermined range from the original guidance path. It may be estimated.

  In the example of the scene shown in FIG. 12, it is estimated that the vehicle 1 is traveling straight and that there is no route (side road) that may be erroneously selected by the driver before or behind the original guidance route. Yes. Therefore, in step S203, the behavior candidate prediction function refers to the table shown in FIG. 9, and scene information in the vehicle state “straight ahead” and the surrounding situation “route guidance left turn instruction (no side road)”, that is, left by the driver. A value “0.7” corresponding to the possibility of selecting movement in the direction and a value “0.1” corresponding to the possibility of selecting movement in the right direction by the driver are acquired. As a result, the behavior candidate prediction function predicts that the vehicle 1 can take one of three behaviors of turning left, turning right, or going straight by a driver's operation.

Next, in the behavior estimation process in step S3, the behavior estimation function of the driving support device 10 obtains and calculates the deviation amount _Xd based on the driver's line-of-sight direction, similarly to the above-described scene example shown in FIG. shift amount X _d and the threshold t _r which _is, by comparing the t _l, estimates the behavior of the vehicle 1 the driver to select.

In the situation where the left turn instruction route guidance information is presented from the navigation device 50 as in the example of the scene shown in FIG. 12, the vehicle 1 is unlikely to turn right by the driver's operation. weights w _r corresponding to the movement (turn right) to have been set as low as "0.1". Therefore, the threshold value t _r for the driver to predicted to turn right the vehicle 1 is calculated as a small value. As a result, the right turn of the vehicle 1 that is unlikely to be selected by the driver can be estimated with a small shift in the driver's line-of-sight direction to the right, so that driving assistance based on the estimation result is performed early. be able to. As a result, it is possible to reduce erroneous selection of the right turn of the vehicle 1 that is unlikely by the driver. Thus, the behavior of the vehicle 1 selected by the driver is appropriately estimated by setting the weight according to the possibility of being selected by the driver to the behavior that the vehicle 1 can take by the operation of the driver. Can do.

  For example, in the situation where the left turn instruction route guidance information is presented by the navigation device 50 as in the example of the scene shown in FIG. 12, when a right turn of the vehicle 1 is predicted from the estimated behavior of the vehicle 1, the driving support device The vehicle controller 40 that has acquired the behavior information from 10 again controls the presentation device (not shown) so as to present the route guidance information for the left turn instruction. Thereby, the driving assistance apparatus 10 can make a driver | operator recognize the route guidance information of a left turn instruction | indication.

  Next, the driving support processing in the driving support device 10 according to the first embodiment, as shown in FIG. 13, in the situation where the route guidance information of the left turn instruction is presented from the navigation device 50, and the scene example shown in FIG. In contrast, an explanation will be given assuming an example of a scene (with a side road) where there is a path (side road) that may be erroneously selected by the driver before or behind the original path. FIG. 13 is a schematic diagram of route guidance information displayed on a display provided in the navigation device 50 in an example of a route guidance left turn instruction (with side road).

  First, in step S1, the gaze direction detection function of the driving assistance device 10 detects coordinate data in the gaze direction of the driver, and stores the detected coordinate data in the RAM provided in the driving assistance device 10, as in the above-described example of the scene. Remember.

  Next, in step S201, the behavior candidate prediction function of the driving assistance device 10 is the vehicle speed, steering angle, etc. acquired from the vehicle controller 40, as in the above-described example of the left turn instruction route guidance (without side roads) shown in FIG. The vehicle 1 is estimated to be in a straight traveling state based on the vehicle information.

  Next, in step S202, the route guidance information for the left turn instruction “turn left on the road ahead” transmitted from the navigation device 50 is acquired by the behavior candidate prediction function. When the route guidance information for the left turn instruction is acquired, the behavior candidate prediction function drives in front of or behind the original guidance route based on the own vehicle position information, map information, and road information acquired from the navigation device 50. It is estimated whether there is a route (side road) that may be selected by the user by mistake. Here, in the example of the scene shown in FIG. 13, there is a route (side road) that may be erroneously selected by the driver before the original guidance route. Based on the information, the map information, and the road information, it is estimated that there is a route (side road) that may be erroneously selected by the driver before or behind the original guidance route.

  In the example of the scene shown in FIG. 13, it is estimated that the vehicle 1 is traveling straight and that there is a route (side road) that may be erroneously selected by the driver before or behind the original guidance route. Yes. Therefore, in step S203, the behavior candidate prediction function refers to the table shown in FIG. 9, and scene information in the vehicle state “straight ahead” and the surrounding situation “route guidance left turn instruction (with side road)”, that is, left by the driver. A value “0.5” corresponding to the possibility of selecting movement in the direction and a value “0.1” corresponding to the possibility of selecting movement in the right direction by the driver are acquired. As a result, the behavior candidate prediction function predicts that the vehicle 1 can take one of three behaviors of turning left, turning right, or going straight by a driver's operation.

Here, in the situation where the left turn instruction route guidance information is presented from the navigation device 50, if there is a route (side road) that may be erroneously selected by the driver, the driver may select it incorrectly. Compared to the case where there is no characteristic route (side road), the possibility that the vehicle 1 turns left on the original guidance route by the operation of the driver is reduced. Therefore, as shown in FIG. 9, when the vehicle state is “straight ahead” and the surrounding situation is “left turn instruction route guidance (with side road)”, the left turn instruction route guidance (without side road) is the weights w _l corresponding to the movement in the direction (left) is set as low as "0.5", the threshold value t _r for the driver to predicted to left of the vehicle 1 is also left instruction route guidance (without side street) It is calculated as a smaller value than in the case of. As a result, in the case of left turn instruction route guidance (with side roads), the left turn of the vehicle 1 by the driver's operation is relatively small deviation in the driver's line of sight compared to the case of left turn instruction route guidance (without side roads). Therefore, driving support based on the estimation result can be performed at an early stage. Thereby, it is possible to reduce the driver's erroneous selection of a route (side road) different from the original guidance route.

In the behavior estimation process in step S3, the behavior estimation function obtains the deviation amount _Xd based on the driver's line-of-sight direction, and calculates the deviation amount _Xd and the threshold _tr , The behavior of the vehicle 1 selected by the driver is estimated by comparing with t _l .

  For example, in the situation where the route guidance information is presented by the navigation device 50 to turn left at the next intersection as in the example of the scene shown in FIG. 13, the vehicle 1 determines the intersection at the front from the estimated behavior of the vehicle 1. When it is predicted that the vehicle will turn left, the vehicle controller 40 that has acquired the behavior information from the driving support device 10 uses a presentation device (not shown) to perform voice guidance such as “Please turn left at a slightly further intersection”. Control. Thereby, the driving assistance device 10 can assist the driving of the driver so that the driver moves along the guidance route.

  In the example of the scene shown in FIG. 13, when estimating the surrounding situation in step S202, whether there is a route (side road) that may be erroneously selected by the driver only in front of the original guidance route. You may make it estimate. In this case, when the vehicle 1 is traveling in front of a route (side road) that may be erroneously selected by the driver, in step S202, the driver is mistakenly in front of the original guidance route. Since it is presumed that there is a route (side road) that may be selected, in step S203, scene information in the vehicle state “straight ahead” and the surrounding situation “route guidance left turn instruction (with side road)”, that is, the driver Thus, the value “0.5” corresponding to the possibility of selecting the leftward movement and the value “0.1” corresponding to the possibility of the rightward movement being selected by the driver are acquired. The On the other hand, when the vehicle 1 crosses a path (side road) that may be erroneously selected by the driver (before the original guidance path in FIG. 13, there is a possibility of being erroneously selected by the driver). At the time of traveling in the back of a certain side road), in step S202, it is estimated that there is no route (side road) that may be erroneously selected by the driver before the original guidance route. In step S203, the scene information in the vehicle state “straight ahead” and the surrounding situation “route guidance left turn instruction (no side street)”, that is, a value “0” corresponding to the possibility that the driver selects the leftward movement. .7 ”and the value“ 0.1 ”corresponding to the possibility of the rightward movement being selected by the driver.

  As described above, in the first embodiment, since the behavior selected by the driver is estimated based on the shift in the driver's gaze direction, a small change in the driver's gaze direction can be detected. Among the series of actions for selecting the behavior of the vehicle 1, the behavior of the vehicle 1 can be estimated at an early stage when the movement in the line-of-sight direction is small. In particular, in the first embodiment, a plurality of behaviors that the vehicle 1 can take by the driver's operation in the estimated vehicle state / ambient situation are predicted, and the behavior of the vehicle 1 that the driver selects from among the plurality of predicted behaviors is predicted. presume. Therefore, it is possible to identify a scene where the driver needs to be selected, in particular, a scene where the driver is confused, and to estimate the behavior of the vehicle 1 selected by the driver. Even when the behavior of the vehicle 1 selected by the driver based on the small movement in the direction is estimated, the behavior of the vehicle 1 can be appropriately estimated.

  In addition, according to the present embodiment, each of a plurality of behaviors predicted to be taken by the vehicle 1 by the driver's operation for each vehicle state and surrounding situation is weighted according to the possibility that each behavior is selected. It is set, and a threshold for estimating the behavior of the vehicle 1 selected by the driver is calculated using the weight set for each behavior. In particular, according to the present embodiment, the weight corresponding to the behavior that is unlikely to be selected is set to a small value. Therefore, it is possible to estimate the behavior of the vehicle 1 that is unlikely to be selected even at an early stage when the driver's line-of-sight direction shift is small, and early control and information presentation of the vehicle 1 according to the estimated behavior are performed. Yes. As a result, the behavior that is unlikely to be selected can be reduced from being erroneously selected by the driver.

ここで、算出される偏移量ｌは正の値となるため、挙動推定機能は、角度θを用いて、算出された偏移量ｌが何れの方向を示しているかを区別する。例えば、角度θが−４５°〜＋４５°である場合には右方向、角度θが＋４５°〜＋１３５°である場合には上方向、角度θが＋１３５°〜―１３５°である場合には左方向、角度θが−１３５°〜−４５°である場合には下方向への偏移を示していると判断する。 In the present embodiment, the behavior of the vehicle 1 selected by the driver is estimated using only the lateral direction (X-axis direction) shift amount _Xd, but the lateral direction (X-axis direction) is estimated. also used shift amount Y _d in the vertical direction in addition to the shift amount X _d (Y-axis direction), it may be estimated behavior of the vehicle 1 the driver to select. Specifically, the behavior estimation function obtains the shift amount l and the angle θ based on the following formulas (7) and (8).

Here, since the calculated shift amount l is a positive value, the behavior estimation function uses the angle θ to distinguish which direction the calculated shift amount l indicates. For example, when the angle θ is −45 ° to + 45 °, rightward, when the angle θ is + 45 ° to + 135 °, upward, and when the angle θ is + 135 ° to −135 °, left When the direction and angle θ are −135 ° to −45 °, it is determined that a downward shift is indicated.

ここで、変化幅ｌ_ｓは、偏移量ｌが採りうる変化幅であり、実験などにより設定される任意の値である。これにより、挙動推定機能は、算出した角度θが−４５°〜＋４５°までの角度である場合には、偏移量ｌと右方向に対応する閾値ｔ_ｒとを比較して、偏移量ｌが閾値ｔ_ｒを超える場合には、車両１が右車線に車線変更するものと推定する。同様に、挙動推定機能は、算出した角度θが＋１３５°〜―１３５°までの角度である場合には、偏移量ｌと左方向に対応する閾値ｔ_ｌとを比較して、偏移量ｌが閾値ｔ_ｌを超えた場合には、車両１が左車線に車線変更するものと推定する。 In this case, the behavior estimation function is calculated on the basis of the threshold value t _r, the t _l the following equation.

Here, the change width l _s is a change width that can be taken by the shift amount l, and is an arbitrary value set by an experiment or the like. Thus, the behavior estimation function, when the calculated angle θ is an angle of up to -45 ° ~ + 45 ° compares the threshold value t _r corresponding to the shift amount l and to the right, shift amount If l exceeds the threshold _tr , it is estimated that the vehicle 1 changes to the right lane. Similarly, when the calculated angle θ is an angle from + 135 ° to −135 °, the behavior estimation function compares the deviation amount l with the threshold value t ₁ corresponding to the left direction, and the deviation amount is calculated. when l is greater than the threshold value t _l is estimated that the vehicle 1 is lane change to the left lane.

<< Second Embodiment >>
Next, the second embodiment will be described. In the second embodiment, in the vehicle 1 similar to that in the first embodiment shown in FIG. 1, as described below, the driving support process in the driving support device 10 is performed according to the flowchart shown in FIG. 14. Here, FIG. 14 is a flowchart showing a driving support process in the driving support apparatus 10 according to the second embodiment. In the following, as in the first embodiment, the driver A is approaching the preceding vehicle, and the vehicle 1 travels in the second lane of traffic on the three-lane road (central lane). A description will be given assuming the scene example shown in FIG. Note that the driving support processing according to the second embodiment is also executed by the driving support device 10 as in the first embodiment.

  First, in step S10, the gaze direction of the driver is detected by the gaze direction detection function as in step S1 of the first embodiment.

  In step S20, as in step S2 of the first embodiment, the vehicle state and the surrounding situation are estimated by the behavior candidate prediction function of the driving support device 10. In the estimated vehicle state and the surrounding situation, the driver's A plurality of behaviors that the vehicle 1 can take are predicted by the operation.

  Next, in step S <b> 21, the driver is identified based on the image information transmitted from the camera 30 by the driver identification function of the driving support device 10. In the present embodiment, the driver is identified as the driver A by the driver identification function.

  Next, the behavior estimation process in step S30 will be described. In step S30, in addition to the coordinate data in the driver's line-of-sight direction, a plurality of behaviors predicted to be taken by the vehicle 1 by the driver's operation based on the driver information of the identified driver and the behavior ratio described later. The behavior selected by the driver is estimated. Here, FIG. 15 is a flowchart showing the behavior estimation processing in step S30 according to the second embodiment. The contents of the behavior estimation process in step S30 will be described below along the flowchart of FIG.

First, in step S311, as in step S301 of the first embodiment, the center of gravity of the distribution of the coordinate data (X _i , Y _i ) of the gaze direction for a certain time detected by the gaze direction detection function by the behavior estimation function. Coordinates (X _c , Y _c ) are calculated.

In step S312, the centroid coordinates (X _c , Y _c ) at the time when a plurality of behaviors are predicted by the behavior candidate prediction function are the initial centroids (X ₀ , Y ₀ ), as in step S302 of the first embodiment. Is calculated as The initial centroids _(X 0, _{Y 0)} in the post-calculation of the initial centroid calculation after the sight line direction of the coordinate data of the center of gravity coordinates and the initial center of gravity _(X 0, _{Y 0)} difference deviation amount between the _X d, _Calculated as Yd.

ここで、挙動割合ｐ_ｒ，ｐ_ｌは、運転者により選択される車両１の挙動に対応しており、推定した車両状態および周囲状況において、識別された運転者が過去に選択した車両１の挙動の割合に応じて算出される値である。 Next, in step S313, the behavior estimation function, the calculated shift amount X _d and the threshold value t _r, and the t _l are compared, the behavior of the vehicle 1 the driver selects is estimated. However, in the second embodiment, unlike the first embodiment, the behavior estimation function, the following equation (11), calculates a threshold value _t r, _{t l} based on (12).

Here, the behavior ratios p _r and p ₁ correspond to the behavior of the vehicle 1 selected by the driver, and the vehicle 1 selected by the identified driver in the past in the estimated vehicle state and surrounding situation. It is a value calculated according to the behavior ratio.

Here, a method of calculating the behavior ratios p _r and p ₁ will be described with reference to the flowchart shown in FIG. FIG. 16 is a flowchart showing the behavior ratio calculation process. The behavior ratio calculation process is performed at the timing when the behavior of the vehicle 1 is selected by the driver, separately from the driving support process by the driving support device 10 illustrated in FIG.

  As shown in FIG. 16, in step S40, the driving history learning function of the driving support device 10 is based on the scene information, the vehicle information, and the surrounding information, and the vehicle 1 selected by the driver in the estimated vehicle state and surrounding conditions. The driver's selection result for each vehicle state and surrounding situation is acquired as a driving history. Specifically, the driving history learning function includes vehicle information such as a turn signal lighting signal obtained from the vehicle controller 40, a steering angle by a steering operation, or a brake signal, and vehicle 1 position information obtained from the navigation device 50. And the behavior of the vehicle 1 selected by the driver is determined based on the surrounding information such as the vehicle traveling direction. For example, when the driver changes the lane of the vehicle 1, generally, the steering operation is performed following the operation of the direction indicator by the driver. Therefore, by recognizing this series of operations as a pattern, the vehicle 1 It may be determined whether or not the vehicle has changed lanes. Further, based on the road information acquired from the navigation device 50 and the traveling direction of the vehicle 1, it may be determined whether the vehicle 1 has made a right or left turn or a lane change.

Next, in step S50, the driving history learning function stores the driving history acquired in step S40 in the RAM of the driving support device 10 in association with the driver information of the driver identified in step S20. The driving history learning function may store the shift amount _Xd when the driver selects the behavior of the vehicle 1 in association with the driving history. Thereby, for example, when calculating the behavior ratio described later, it is possible to calculate the behavior ratio in consideration of the shift amount _Xd when the driver selects the behavior of the vehicle 1.

Furthermore, in step S50, the driving history learning function totals past driving histories including the driving history acquired in step S40, and each behavior that the identified driver has selected in the past in the estimated vehicle state and surrounding situation. Calculate the percentage of. Here, FIG. 17 is a diagram illustrating an example of a table in which scene information according to the second embodiment is recorded. The driving history learning function multiplies the calculated behavior ratios p _r and p _l by the basic weights w _r and w _l , respectively, and weights w _r p _r and w _{l in} consideration of the behavior ratios p _r and p _l for each driver. a p _l be set in the table of FIG. 17. Thus, in step S313, the weights w _r and w _l and the behavior ratios p _r and p _l can be acquired by referring to the table shown in FIG. Note that the behavior ratios p _r and p _l are the ratios of behaviors selected by a predetermined driver in a predetermined vehicle state and surrounding situation, and therefore the behavior ratios p _r and p _l are considered in the table shown in FIG. The weights w _r p _r and w _l p _l are set for each vehicle state / ambient situation and for each driver.

For example, in the example of the scene shown in FIG. 4, the vehicle 1 driven by the identified driver A is traveling on the second traffic zone (central lane) of the road with three lanes on one side while approaching the preceding vehicle. In the situation where the vehicle 1 is in an approaching state with the preceding vehicle and the vehicle 1 is traveling on the second lane (central lane) of the road with three lanes on one side, Calculates the behavior ratios p _r and p _l by summing up the operation history of each behavior selected in the past. Then, the behavior estimation function is a weight w obtained by multiplying the basic weights w _r and w _l of the vehicle state “approaching the preceding vehicle” and the surrounding situation “three lanes on one side, second lane” by the behavior ratios p _r and p _{l. r} p _r and w _l p _l are set in the table of FIG. Here, in the vehicle state “approaching the preceding vehicle” and the surrounding situation “three lanes on one side, second lane”, the number of times the driver A changed the lane of the vehicle 1 to the right lane changes the lane of the vehicle 1 to the left lane. If greater than the number of times the behavior ratio p _r corresponding to the right direction is calculated as the behavior percentage p _l a value greater than that corresponding to the left direction. Accordingly, in the table of FIG. 17, the weight (w _r p _r ) corresponding to the right direction is set to “0.8” that is larger than the basic weight w _r “0.5”, for example, The weight corresponding to the direction (w _l p _l ) is set to “0.2” which is smaller than the basic weight w _l “0.5”.

Incidentally, as in this embodiment, the ratio p _r behavior selected by the _driver, the threshold value based on p _l t _r, when calculating the t _l, driver information of the behavior of the vehicle 1 previously selected If not accumulated to some extent, there is a case where the threshold t _r, t _l is unable to reflect the proper tendency of the driver, which may give uncomfortable feeling to the driver to perform the driving assistance based on such threshold. Therefore, it is preferable to set the behavior ratios p _r and p ₁ to 1 until a predetermined number or more of operation histories are accumulated. Thus, until the predetermined number or more of driving histories are accumulated, the behavior ratios p _r and p _l are not reflected in the threshold values _tr and t _l , and at least the same effects as those of the first embodiment can be achieved.

  After estimating the behavior of the vehicle 1 selected by the driver in step S313, the driving support process by the driving support device 10 is terminated. As in the first embodiment, the driving support device 10 transmits the estimated behavior information of the vehicle 1 to the vehicle controller 40, and the vehicle controller 40 that has acquired the behavior information performs a preparatory operation in preparation for the driver's operation. In this way, the vehicle 1 is controlled, or information is presented to the driver at an early stage to assist the driver in driving. Further, in the present embodiment, by using the driver information of the driver identified by the driver identification function, information based on the behavior of the vehicle 1 estimated at the timing, frequency, or method set by the driver is obtained. Can present.

As described above, in the second embodiment, the behavior estimation function is the driving history for each driver, that is, the ratios p _r and p of the behaviors selected by each driver in the past in the estimated vehicle state and surrounding situation. weight in consideration of the _{l (w r p r, w} l p l) calculating a threshold value _t r, _{t l} with. Then, the calculated threshold value _t r, compared with the shift amount _{X d} of _{t l} and the transverse direction (X-axis direction), if the shift amount _{X d} exceeds either threshold _t r, _{t l} is exceeded vehicle 1 is assumed to move in a direction corresponding to the threshold was, and when the shift amount X _d does not exceed the threshold value t _r, t _l, the vehicle 1 is estimated to be straight. Thus, in the second embodiment, the weights w _r, in addition to w _l, the proportion of behavior the driver identified in the estimated vehicle condition and the surrounding condition has selected in the past p _r, based on the p _l By calculating the thresholds t _r and t _l , the behavior of the vehicle 1 with a low ratio selected by the driver in the past can be estimated even at an early stage when the shift of the driver's line-of-sight direction is small. Control of the vehicle 1 and information presentation according to can be performed at an early stage. As a result, it is possible to reduce the fact that a behavior with a low ratio selected by the identified driver in the past is erroneously selected by the driver. Thus, in 2nd Embodiment, in addition to the effect of 1st Embodiment, according to a driver | operator's tendency, the behavior of the vehicle 1 which a driver | operator selects can be estimated appropriately.

  The embodiment described above is described for facilitating understanding of the present invention, and is not described for limiting the present invention. Therefore, each element disclosed in the above embodiment is intended to include all design changes and equivalents belonging to the technical scope of the present invention.

  That is, the present invention is not limited to the above-described embodiments, and the embodiments described above may be combined.

For example, in the second embodiment, the ratios p _r and p _l of the behavior selected by the driver in the past in the estimated vehicle state and surrounding situation are obtained. However, the present invention is not limited to this, and the vehicle is not identified. The behavior ratios p _r and p ₁ may be obtained based on the driving history of the driver who drove 1. Further, the driver may be able to set the behavior ratios p _r and p _l .

Note that the gaze direction detection function of the above-described embodiment is the gaze direction detection means of the present invention , the vehicle controller 40 is the vehicle information acquisition means and control means of the present invention, and the vehicle controller 40 and the navigation device 50 are the ambient information of the present invention. In the acquisition means, the behavior candidate prediction function is in the behavior candidate prediction means of the present invention, the behavior estimation function is in the deviation amount calculation means and behavior estimation means of the present invention, the driver identification function is in the driver identification means of the present invention, The driving history learning function corresponds to the storage means of the present invention.

DESCRIPTION OF SYMBOLS 1 ... Vehicle 10 ... Driving assistance device 20 ... Infrared illumination 30 ... Camera 40 ... Vehicle controller 50 ... Navigation device

Claims (5)

Gaze direction detecting means for repeatedly detecting the gaze direction of the driver;
Vehicle information acquisition means for acquiring vehicle information;
Ambient information acquisition means for acquiring information around the vehicle;
Behavior candidate prediction means for predicting a plurality of behaviors that can be taken by the vehicle by a driver's operation based on the vehicle information and information around the vehicle;
Based on the driver's line-of-sight direction detected by the driver's line-of-sight direction, the center of gravity of the driver's line-of-sight distribution is repeatedly calculated, and the calculated center of gravity of the driver's line-of-sight direction is preset. A deviation amount calculating means for repeatedly calculating a difference in the vehicle width direction from the predetermined initial center of gravity as a deviation amount in the vehicle width direction;
By determining whether or not a deviation amount in the vehicle width direction exceeds a threshold corresponding to a predetermined behavior among the plurality of predicted behaviors, the behaviors predicted by the behavior candidate prediction unit are determined. A behavior estimation means for estimating the behavior selected by the driver,
And a control means for performing vehicle control and / or information presentation based on the behavior estimated by the behavior estimation means. The driving support device according to claim 1,
The shift amount calculating means sets the center of gravity of the distribution in the gaze direction of the driver detected in advance for a predetermined time as the initial center of gravity. The driving support device according to claim 1 or 2,
The threshold value corresponding to the predetermined behavior is set according to the selectability of the behavior. The driving support device according to any one of claims 1 to 3,
A driver identification means for identifying the driver;
The threshold value corresponding to the predetermined behavior is set according to the driver identified by the driver identifying means. The driving support device according to claim 4,
It further has storage means for storing the driving history of the driver for each driver,
The threshold value corresponding to the predetermined behavior is set based on the driving history of the driver stored by the storage unit.

Images