JP4901287B2 - Driving support device - Google Patents

Description

  The present invention relates to a driving support device, and more particularly to a driving support device that detects a situation in which an accident is likely to occur during driving of a vehicle and performs an operation such as urging a driver to pay attention.

  Situation judgments and confirmation of risk factors when driving a vehicle such as an automobile or a truck depend solely on the driver's own attention and observation. For this reason, attention may be distracted depending on the physical condition and psychological state of the driver, resulting in oversight of pedestrians, bicycles, and other vehicles. In the worst case, collisions with pedestrians may occur. May lead to a connection.

  In order to avoid such oversight of pedestrians and the occurrence of collision accidents, various driving support devices that support safe driving by judging the possibility of collision with pedestrians and the like have been developed.

  For example, using a so-called car navigation system, based on the map data etc., evaluate whether the intersection is likely to make a driving mistake from the width and type of the intersection, the connection relationship of the road at the intersection, etc. There is known a driving support device configured to emit an alarm sound before an intersection determined to be important (see Patent Document 1).

  In addition, it is equipped with an accident frequent occurrence point database that stores information on the latitude and longitude of locations where accidents frequently occur from the police and information centers, etc., and information on the type of accident, etc., when the vehicle approaches the accident frequent occurrence point, A driving support device that reads out a caution code from a database and emits a characteristic alarm sound has been proposed (see Patent Document 2).

In addition, danger information in the event that the driver is near or disappointed during driving is automatically collected and stored from the driver and residents as information including the location where the dangerous situation occurred and the type of danger. A danger information collection and distribution device that distributes danger information in response to a request from an in-vehicle terminal has been proposed (see Patent Document 3). The in-vehicle terminal emits a warning sound immediately before the dangerous area or while passing through the dangerous area.
JP-A-9-190596 JP 2002-90154 A JP 2003-123185 A

  However, in these driving support devices and danger information collection and distribution devices, the driver pays sufficient attention and checks at intersections, accident-prone points, and dangerous areas that are determined to be cautioned based on the shape of the intersection itself. Even if there is a warning sound, etc., it is performed for each dangerous area.

  For this reason, the driver feels bothered to operate the device so that the warning sound is not produced, or the driver gets used to the warning sound and cannot draw attention, and the warning is meaningful. There is a risk that it will disappear.

  In order to effectively avoid collisions with pedestrians and effectively support safe driving, avoid collisions by focusing on the drivers themselves rather than on the basis of frequent accidents such as those described above. Etc. are necessary. That is, although the current driving state of the host vehicle and the surrounding environment of the host vehicle are the same driving state as when the driver had caused an accident in the past or is about to cause an accident, the driver It is considered that it is more effective to call attention or take a collision avoidance measure when performing the same operation as before.

  The present invention has been made in view of the above circumstances, and accurately supports driving when the driver performs the same operation in a situation similar to the situation in which the accident has occurred or is likely to cause an accident in the past. An object of the present invention is to provide a driving support device capable of performing the above.

In order to solve the above-mentioned problem, the invention according to claim 1
In a driving support device that determines the possibility of a collision of the host vehicle and supports driving of the vehicle,
A detection device for detecting data relating to the running state of the host vehicle and data relating to the surrounding environment of the host vehicle ;
Determination means for determining whether or not the driving state and the surrounding environment have a high possibility of collision based on the data detected by the detection device;
When the determination unit determines that the driving state and the surrounding environment have a high possibility of a collision, the data on the driving state of the own vehicle and the data on the surrounding environment of the own vehicle at a time point that is a predetermined time later than the predetermined state the data obtained from Step a storage device for storing the data of the past danger information in the vehicle,
A correlation value indicating a correlation state between the data relating to the traveling state of the own vehicle at the time of the current sampling detected by the detection device and the data relating to the surrounding environment of the own vehicle and the data of the past danger information stored in the storage device; calculated, the correlation value calculated is equal to or obtaining Bei a determination means for determining whether the possibility of collision at the time of the current sampling by whether exceeds a preset threshold.

According to the first aspect of the present invention, the detection device of the driving support device detects data related to the traveling state of the host vehicle such as the steering angle of the steering wheel, the vehicle speed, the accelerator opening, and the brake hydraulic pressure, and the storage device stores the host vehicle. In this case, data relating to the traveling state of the host vehicle such as the steering angle of the steering wheel when an event such as sudden braking has occurred in the past and there is a possibility of a collision is stored. In addition, the judging means collates the data input from the current detecting means with the data of past danger information stored in the storage device, calculates their correlation value, and the correlation value is high. If it exceeds a preset threshold value, it is determined that there is a possibility of collision. In addition to data indicating the driving state of the host vehicle, the data detected by the driving support device and used for determination and judgment, road shape information such as whether the vehicle is an intersection, whether it is straight or curved, and urban or suburban Data relating to the environment around the host vehicle including information on such areas is also used.

  According to a second aspect of the present invention, in the driving support apparatus according to the first aspect, when the determination unit determines that there is a possibility of a collision based on the correlation value, a predetermined response unit of the host vehicle is provided. Is operated.

  According to the second aspect of the present invention, when the determining means determines that the correlation value is large and the degree of correlation between the current data and the past risk information data is high, an alarm device or assist device for the own vehicle is provided. Driving support is performed by operating a response unit such as a control device.

The invention according to claim 3 is the driving support device according to claim 1 ,
The determination means calculates a feature amount from data detected by the detection device when it is determined that the traveling state has a high possibility of a collision, and the feature amount is used as past risk information data in the host vehicle. Storing in the storage device;
The determination unit calculates a feature amount from data detected by the detection device, and calculates a correlation value between the calculated feature amount and the feature amount as data of the past danger information stored in the storage device. It is characterized by that.

According to a third aspect of the present invention, the determination means stores feature data that reflects a tendency of data instead of the raw data detected by the detection device when storing past risk information data in the host vehicle. Compressed into memory. In addition, the judging means calculates a feature amount from the current data, calculates a correlation value with risk information data as a past feature amount, the degree of the correlation is high, and the correlation value exceeds a preset threshold value. If there is a collision, it is determined that there is a possibility of a collision.

  According to the first aspect of the present invention, when the correlation value is a value close to the maximum value and the degree of correlation is high, the current running state of the host vehicle is likely to cause an accident or an accident in the past. The situation is similar to the situation described above, and it can be seen that the driver is performing the same operation as in the past, so it is possible to provide driving assistance effectively and accurately by calculating and determining the correlation value. It becomes.

In addition, past dangerous information data is accumulated as data up to several seconds before an event that may have caused a collision, and by correlating it with the current data, several seconds when an event that has the possibility of a collision occurs. It becomes possible to determine whether or not there is a high possibility of collision before. Therefore, it is possible to provide driving support with a sufficient margin.
Also, as data used for determination and judgment detected by the driving support device, not only data indicating the driving state of the host vehicle but also data related to the surrounding environment of the host vehicle including road shape information, area information, and the like are used. Thus, it is possible to grasp in what surrounding environment and in what running state a past event that may have caused a collision. Moreover, it is possible to accurately determine whether or not the current running state and the surrounding environment are similar to the past situation, and the effects of the invention described in the above claims can be more reliably exhibited, and the device It becomes possible to further improve the reliability of.
In addition, the determination means determines that the traveling state has a high possibility of a collision when, for example, sudden braking is performed, and accumulates each data at that time in a storage device. Data is accumulated not as a frequent accident point, but as past danger information in the vehicle. Therefore, it is possible to perform driving support such as accurate collision avoidance according to the tendency or tendency of the driver to overlook obstacles and pedestrians.

  According to the second aspect of the present invention, when it is determined that the correlation value is large and the current running state collides with an obstacle, a pedestrian, or the like as in the past event, causes an accident or is likely to collide. For example, the driving support device activates an alarm device of the own vehicle to alert the driver, or operates the assist control device to perform an operation for avoiding a collision. Therefore, the effect of the said Claim 1 is exhibited more correctly.

According to the invention described in claim 3 , the information such as the steering angle and speed of the steering wheel, the accelerator opening degree, and the brake hydraulic pressure representing the running state is not data only at the present time, but at a certain time from before. More accurate driving support can be performed by using sequential data.

  At that time, by extracting feature values from the respective time-series data of the current time-series data and the time-series data as past danger information, the current situation is determined as a collision. Therefore, it is possible to clarify whether and how much the situation is similar to the past situation, and it is possible to improve the reliability and robustness of the apparatus.

  Further, if the feature amount is extracted as in the present embodiment, the data amount can be reduced, and the search and calculation can be performed in a shorter time, and the effects of the invention described in the above claims can be achieved. It can be exhibited more effectively.

  DESCRIPTION OF EMBODIMENTS Hereinafter, an embodiment of a driving support apparatus according to the present invention will be described with reference to the drawings.

  The driving support apparatus according to the present embodiment is configured by a computer in which a CPU, a ROM, a RAM, an input / output interface and the like (not shown) are connected by a bus. In addition, it is also possible to integrate the driving assistance apparatus of this embodiment into ECU (Electric Control Unit or Engine Control Unit), for example, and to be integrated with ECU.

  As illustrated in FIG. 1, the travel support device 1 includes a detection device 2, a determination unit 3, a storage device 4, and a determination unit 5. In the present embodiment, the detection device 2 includes various sensors 6, a navigation device 7, and an ambient environment recognition device 8.

  The travel support apparatus 1 of the present embodiment includes a vehicle speed sensor that detects a vehicle speed as various sensors 6, a hydraulic pressure sensor that detects a brake operation, a rudder angle sensor that detects a steering angle, an accelerator opening sensor that detects an accelerator operation, An acceleration sensor for detecting the acceleration of the host vehicle is provided, and light switch operation information, turn signal switch operation information, wiper switch operation information, brake operation information, side brake operation information, and time information are input. Yes.

  These measurement values and information are transmitted from the various sensors 6 to the determination means 3. As time information, date, time, season, day and night are input. It is also possible to configure so that measurement values and information from sensors other than these may be input.

  In addition, the navigation device 7 has a GPS (Global Positioning System) receiver and the like, and can acquire position information of the own vehicle. The road shape of an intersection, straight line or curve, number of lanes, general road or highway A known device having map information that can discriminate between urban areas, suburbs, and mountains is used. These pieces of information are transmitted from the navigation device 7 to the determination means 3.

  The surrounding environment recognition device 8 captures the surrounding environment of the vehicle with an imaging device such as a CCD (Charge Coupled Device) camera, detects the distance to the white line on the road surface, the presence / absence of a signal, and the presence / absence of a wall. It is a device for detecting the distance to obstacles such as other vehicles and the distance to pedestrians. In the present embodiment, the white line on the road surface means a sign represented by a straight line or a broken line such as white or yellow that divides a lane such as a lane marking or a vehicle traffic zone boundary line.

  In the present embodiment, the surrounding environment recognition device 8 uses the three-dimensional object monitoring device described in Japanese Patent Application No. 2004-252521 filed by the applicant of the present application. Here, the three-dimensional object monitoring apparatus will be briefly described.

  The three-dimensional object monitoring device 9 includes a distance measuring device 10 and an image processing device 14 as shown in FIG.

  The distance measuring device 10 includes two imaging devices 11a and 11b, such as a CCD camera, separated by a certain distance in the horizontal direction, an image processor 12, and a memory 13. In the image processor 12, the parallax for the same three-dimensional object is calculated based on a pair of captured images P as illustrated in FIG. 3 captured by the imaging devices 11a and 11b, and the calculated parallax is shown in FIG. Such a distance image D is sent to the memory 13 and stored therein. Note that FIG. 3 illustrates a reference captured image captured by the imaging device 11a among the pair of captured images.

  A distance image D is input from the memory 13 of the distance measuring device 10 to the road shape detecting unit 15 and the three-dimensional detecting unit 16 of the image processing device 14 via the interface circuit 17, and the I / O Measurement values are input from the vehicle speed sensor 21 and the steering angle sensor 22 described above via the interface circuit 18. Further, the image processing apparatus 14 includes a memory 19 for temporarily storing data being processed and an output unit 20 for outputting processed data to the determination unit 3.

  In the image processing apparatus 14, the distance in the real space from the own vehicle to the three-dimensional object is obtained according to the following equations (1) to (3) by using the parallax.

  That is, as shown in FIG. 3, the horizontal axis is the i coordinate axis and the vertical direction is the j coordinate axis with the lower left corner of the distance image D as the origin, and the parallax corresponding to the point (i, j) on the distance image D is expressed as dpij. Further, the three-dimensional coordinate system in the real space is a coordinate system fixed to the host vehicle, the X axis is the lateral direction (right direction is positive) toward the traveling direction of the vehicle, and the Y axis is the vertical direction of the vehicle (up direction is Positive), the Z-axis is the front-rear direction of the vehicle (the forward direction is positive), and the origin is a point on the road surface directly below the center of the two imaging devices 11a, 11b.

In this case, in the image processing device 14, coordinate conversion from the point (i, j) and the parallax dpij on the distance image to the point (x, y, z) on the real space is performed by the following equations (1) to (3). Based on the above.
x = CD / 2 + z * PW * (i-IV) (1)
y = CH + z * PW * (j-JV) (2)
z = CD / (PW × dpij) (3)

  Here, CD is the distance between the two imaging devices 11a and 11b, PW is the viewing angle per pixel, CH is the mounting height of the two imaging devices 11a and 11b, and IV and JV are infinity at the front of the vehicle. The i and j coordinates on the point distance image are represented.

The road shape detection unit 15 performs coordinate conversion of the point (i, j) and the parallax dpij on the distance image D to a point (x, y, z) on the real space, and extracts a white line on the road therefrom. The white line on the extracted road is approximated by a straight broken line expressed by the following equations (4) and (5) to recognize the road shape. The obtained linear road shape parameters a, b, c, d are transmitted to the memory 19 and stored.
x = a × z + b (4)
y = c × z + d (5)

  Further, the road shape detection unit 15 calculates the distance from the vehicle body center of the host vehicle to the right white line and the distance to the left white line based on the equations (4) and (5), and calculates the calculated distances. This information is transmitted to the determination means 3.

  In the three-dimensional object detection unit 16, the distance image D is divided into each section C at a predetermined interval in the horizontal direction as shown in FIG. 5, and the road detected by the road shape detection unit 15 among the points included in each section C. A histogram of distances in real space is created for points above the road surface of the shape, the mode value is calculated, and a three-dimensional object having the mode value as the distance from the host vehicle is extracted for each section C It has come to be.

  Further, in the three-dimensional object detection unit 16, the points where the positions of the three-dimensional objects detected for each section are close to each other are grouped into one group, and as shown in FIG. A solid object is detected as an object O when the lines are arranged substantially in parallel with each other, and as a side wall W when the points are arranged substantially parallel with the traveling direction.

  The three-dimensional object detection unit 16 determines whether there is a signal on the object O and the side wall W on the right side, left side, or both sides of the vehicle, based on the detected shape and size of the object O and the side wall W, the speed with respect to the road surface, and the like. Whether or not is classified and extracted and transmitted to the determination means 3.

  Further, the three-dimensional object detection unit 16 classifies and extracts obstacles and persons such as other vehicles based on the detected object O and the side wall W, and the traveling direction with respect to the own vehicle, that is, the distance in the z-axis direction is determined. For the closest obstacle or person, the travel direction distance and the lateral direction to the obstacle, that is, the distance in the x direction, and the travel direction distance and the lateral distance to the person are calculated, and the information is transmitted to the determination means 3. It is like that.

  Each data shown in the following Table 1 is input to the determination means 3 from the various sensors 6, the navigation device 7, and the surrounding environment recognition device 8 as the detection device 2. In the present embodiment, the determination means 3 is provided with a ring buffer (not shown), and these data transmitted at predetermined time intervals are sequentially recorded in the ring buffer while the write pointer is shifted. Yes.

  In Table 1, when the right white line distance or the left white line distance is 10 m or more, the right white line distance Dr or the left white line distance Dl is 10 m, respectively. When the interpersonal traveling direction distance or the objective traveling direction distance is 100 m or more, The interpersonal travel direction distance Mz or the objective travel direction distance Oz is 100 m. When the interpersonal lateral distance or the objective lateral direction distance is 20 m or more and −20 m or less, the interpersonal lateral distance Mx or the objective lateral direction distance Ox is 20 m, respectively. Or -20 m. The lateral distances Mx and Ox are expressed with a positive value for the right direction and a negative value for the left direction based on the vehicle position.

  Based on the data detected by the detection device 2, the determination unit 3 determines whether the driving state and the surrounding environment have a high possibility of a collision, and the driving state and the surrounding environment have a high possibility of a collision. When the determination is made, the data detected by the detection device 2 is stored in the storage device 4 as past danger information data in the host vehicle.

  In the present embodiment, the determination means 3 determines that the vehicle is in a traveling state and the surrounding environment where the possibility of a collision is high when a sudden brake is stepped on or when a large acceleration is generated in the host vehicle. That is, the determination means 3 is used when the amount of change in the measured value of the brake hydraulic pressure Bp exceeds a preset threshold value, or when the absolute value of the measured value of acceleration a by the acceleration sensor exceeds a preset threshold value. It is determined that the driving state and the surrounding environment have a high possibility of collision.

  When the data detected by the detection device 2 is stored in the storage device 4 as data of past danger information in the host vehicle, the determination means 3 is described in Table 1 two seconds before the occurrence of the event recorded in the ring buffer. Each data of road shape, area information, time information, obstacle information and vehicle switch information is read and converted into a normalized vector form according to the following equations (6) to (10). The data is stored in the storage device 4 as past danger information data.

  Further, for the steering information, speed information, accelerator information, and brake information in Table 1, the determination means 3 reads out data for 5 seconds from 7 seconds to 2 seconds before the occurrence of the event from the ring buffer. Thus, the feature amounts are calculated and stored in the storage device 4 as past risk information data in the host vehicle.

In this embodiment, the determination means 3 reads the steering angle δ data, which is steering information, from the ring buffer for 5 seconds, normalizes it, arranges it in a line on the memory, and fast Fourier transforms the arranged time series data. Transform (FFT: Fast Fourier Transform) is performed, and discrete Fourier transform is performed based on the following equation (11) to obtain a frequency spectrum X _b (i).

Here, x _b (i) is a data string of the steering angle δ for 5 seconds, N is the total number of samples for 5 seconds, Ts is a sampling interval, and i is a value between 0 and N−1. Indicates the component number. The subscript b indicates data of past danger information in the host vehicle.

Subsequently, the determination unit 3 calculates the power spectrum Φ _b (i) from X _b (i) based on the following equation (12).

In this embodiment, the frequency component numbers i = 0, 1,..., N / 2-1 have the four largest frequency component numbers max1, max2, up to the fourth peak value of the power spectrum Φ _b (i). max3 and max4 are extracted as characteristic amounts of the steering angle δ.

  Similarly, the determination unit 3 calculates four feature amounts for the vehicle speed V, the accelerator opening A, and the brake hydraulic pressure Bp, which are speed information, accelerator information, and brake information.

  In this way, the determination means 3 performs the road shape, area information, time information, obstacle information, and obstacle information described in Table 1 for one event determined to be a traveling state and a surrounding environment with a high possibility of collision. Each data converted and vectorized with respect to vehicle switch information and four feature amounts max1, max2, max3, and max4 calculated for steering information, speed information, accelerator information, and brake information are stored in the storage device 4. It is like that.

Further, in the present embodiment, and their data and the feature, the frequency spectrum X _b (max1) which correspond to the four characteristic _{quantity, X b (max2), X} b (max3), X b and (max4) In addition, it is stored as past risk information data for a set of the vehicle. And the determination means 3 is comprised so that the data of the past danger information in one set of own vehicles may be memorize | stored in the memory | storage device 4 whenever the event with high possibility of a collision arises.

  A storage device 4 and a determination unit 5 are connected to the determination unit 3.

  The storage device 4 is composed of, for example, an HDD (Hard Disk Drive) or the like, and a set of feature values transmitted together with the data and frequency spectrum transmitted from the determination means 3 is shown in FIG. Are stored as one case. In FIG. 7, road, area, time, obj, sw, st, v, acc, and brk are the road shape, area information, time information, obstacle information, vehicle switch information, steering information, and speed described in Table 1, respectively. Information, accelerator information and brake information.

  Further, the storage device 4 is also connected to the determination unit 5, and the storage device 4 reads out necessary data for each case in accordance with an instruction from the determination unit 5 and transmits it to the determination unit 5. ing.

  The judging means 5 compares the current data with the past risk information data of the host vehicle stored in the storage device 4, and the current running state of the host vehicle and the surrounding environment are highly likely to collide. It is determined whether or not it is in a state. The comparison between the current data and the past risk information data of the host vehicle is performed by calculating their correlation values.

  The determination means 5 receives the current data of the road shape, area information, time information, obstacle information, and vehicle switch information shown in Table 1 detected by the detection device 2 and input to the determination means 3. It is like that. Based on these data, the judging means 5 converts each data into a normalized vector form according to the equations (6) to (10).

Then, correlation values r _road , r _area , r _time , r _obj between each data vectorized in this way and each of past risk information data in a plurality of sets of own vehicles stored in the storage device 4. , R _sw are calculated according to the following equations (13) to (17). In the following formulas, the subscripts a and b indicate current data and past danger information data of the host vehicle, respectively.

Here, the absolute value of the vector difference represents the Euclidean distance between the vectors. In addition, the correlation values r _road , r _area , r _time , r _obj , and r _sw are closer to the maximum value 1, and the correlation is stronger. Correlation values r _road , r _area , r _time , r _obj , and r _sw are calculated for each set of past risk information data in the vehicle.

The determination means 5 calculates correlation values r _st , r _v , r _acc , r _brk for steering information, speed information, accelerator information, and brake information.

That is, the determination means 5 reads each data of the steering angle δ, the vehicle speed V, the accelerator opening A, and the brake hydraulic pressure Bp for 5 seconds from the present time recorded in the ring buffer of the determination means 3 to 5 seconds before, The frequency spectrum X _a (i) and the power spectrum Φ _a (i) are obtained according to the following equations (18) and (19), respectively. In the following formula (18), x _a (i) represents each data string for 5 seconds from the present time to 5 seconds before.

  Actually, since it takes time to perform the calculation for all the numbers of frequency component numbers i = 0, 1,..., N−1 for the steering angle δ and the like, in this embodiment, the judging means 5 is a storage device. 4 sequentially reads each set of past risk information data stored in the host vehicle, and calculates the formulas (18) and (19) only for the feature amounts max1, max2, max3, and max4 of the set. It is like that.

Further, the determination means 5 uses the conjugate complex number of X _a (maxn) calculated by the equation (18) for the feature amounts max1, max2, max3, and max4 and the frequency corresponding to the feature amount maxn stored in the storage device 4. The cross power spectrum Φ _ab (maxn) is calculated from the spectrum X _b (maxn) according to the following equation (20).

Then, as represented by the following equation (22) below (21) the square value r _n n = 1 to 4 i.e. feature amount of the amplitude value of the coherence Coh _ab (maxn) of formula max1, max2 , respectively calculated for max3, max4, obtains an average of them in accordance with the following equation (23) is adapted to calculate a correlation value r _p.

The correlation value r _p is the correlation value r _st if it is obtained using the feature amounts max1, max2, max3, and max4 with respect to the steering angle δ, and the vehicle speed V, the accelerator opening A, and the brake hydraulic pressure Bp. If found, the correlation values are r _v , r _acc , and r _brk . Further, the correlation values r _st , r _v , r _acc , and r _brk each take a value of 0 or more and 1 or less, and the closer to the maximum value 1, the stronger the correlation. The correlation values r _st , r _v , r _acc , and r _brk are calculated for each set of past risk information data in the _host vehicle.

The judging means 5 uses the weights w _road , preset weights based on the correlation values r _road , r _area , r _time , r _obj , r _sw , r _st , r _v , r _acc , r _brk calculated in this way. Using w _area , w _time , w _obj , w _sw , w _st , w _v , w _acc , w _brk , take their weighted average according to the following equation (24), and calculate it as the correlation value r: It has become.

  The closer the correlation value r is to the maximum value 1, the stronger the correlation between the current data and the past risk information data in the host vehicle.

Correlation values r _road , r _area , r _time , r _obj , r _sw , r _st , r _v , r _acc , and r _brk are calculated for each set of past risk information data in the vehicle. Therefore, the correlation value r is also calculated for each set. That is, correlation values r corresponding to the number of sets of past danger information data in the host vehicle are calculated for each current data detected by the detection device 2 and input to the determination means 3 and the determination means 5. . In the present embodiment, for example, the correlation value r is calculated at intervals of 100 milliseconds.

  Two threshold values rth1 and rth2 (rth1> rth2) are set in advance in the determination means 5 for the correlation value r, and the determination means 5 calculates the running state and surrounding environment of the host vehicle and each past case. If the largest correlation value r among the correlation values r is equal to or greater than the threshold value rth1, the traveling state of the host vehicle and the surrounding environment are classified into level 1 where the possibility of a collision is high. If it is smaller than rth2, the traveling state of the host vehicle and the surrounding environment are classified into level 2 where there is a possibility of collision.

  If the judging means 5 judges that the traveling state of the host vehicle and the surrounding environment are at level 1 or level 2, the responding unit 23 shown in FIG. 1 is actuated to alert the driver or to avoid collision. It is supposed to perform the operation.

  In the present embodiment, when the traveling state of the host vehicle and the surrounding environment are at level 2, the determination unit 5 utters a warning sound indicating that a collision has occurred from the speaker of the warning device 24 that is the responding unit 23. . In addition to this, for example, a buzzer is sounded, a direction to be noticed by the driver based on obstacle information is notified by voice, or an A pillar portion or the like in the direction is illuminated with a lamp, or the direction portion of the bonnet is illuminated. For example, the driver may be alerted.

  Further, when the traveling state of the host vehicle and the surrounding environment are at level 1, the judging means 5 activates the assist control device 25, which is the responding section 23, at the same time as uttering the alarm sound, and automatically applies the brake. Overcoming collision avoidance. In addition to this, for example, it is possible to configure to take measures such as stopping the supply of fuel such as gasoline to the engine.

  Next, the operation of the driving support device 1 according to this embodiment will be described.

  The functions of the components of the driving support device 1 are as described above, and the overall processing flow will be described here.

  Each of the road shape, area information, time information, obstacle information, vehicle switch information, steering information, speed information, accelerator information, and brake information described in Table 1 from the detection device 2 of the driving support device 1 at predetermined sampling intervals. Data is transmitted to the determination means 3. Those data are sequentially recorded in the ring buffer of the determination means 3 and are also transmitted to the determination means 5 via the determination means 3.

  In the determination means 3, when the change amount of the brake hydraulic pressure Bp or the absolute value of the acceleration a exceeds the preset threshold value among the transmitted data, that is, when the sudden braking is stepped on, It is determined that the traveling state and the surrounding environment are in a state where the possibility of collision is high.

Then, the road shape, area information, time information, obstacle information, and vehicle switch information two seconds before that point are read from the ring buffer and vectorized. For steering information, speed information, accelerator information, and brake information, four feature values max1, max2, max3, max4, which are frequency component numbers, respectively, based on data for 5 seconds from 7 seconds to 2 seconds ago. frequency spectra at those frequency components number _{X b (max1), X b} (max2), X b (max3), calculates a X _b (max4). These are combined and stored in the storage device 4 as a set.

  In this way, every time a case with a high possibility of a collision occurs, the determination unit 3 stores the past danger information data in the own vehicle, that is, the so-called near-miss information generated in the own vehicle in the storage device 4. .

  On the other hand, the determination means 5 calculates the feature amount based on the vectorization of the transmitted current data and the data for 5 seconds from the present recorded in the ring buffer of the determination means 3 for 5 seconds. Do it constantly at intervals. Then, each time, the correlation value r is calculated by comparing with the past risk information data of the host vehicle stored in the storage device 4.

  Then, comparing the correlation value r with the threshold values rth1 and rth2, whether or not the current running state of the host vehicle and the surrounding environment are in a state or environment where there is a possibility of a collision, or if there is a possibility of a collision Then it is judged how much possibility.

  Furthermore, when the traveling state of the host vehicle and the surrounding environment are a state or environment where there is a possibility of collision, the determination unit 5 operates the responding unit 23 such as the alarm device 24 or the assist control device 25 to operate the driver. Calls attention to or causes an action to avoid a collision.

  As described above, according to the driving support device 1 according to the present embodiment, the current driving state and the surrounding environment of the host vehicle and the case where there is a high possibility of a collision by actually stepping on the brake in the past. The correlation between the running state and the surrounding environment is calculated as a correlation value r, and the correlation value r is classified by a threshold value to determine whether or not there is a possibility of a collision.

  Therefore, if the correlation value r is a value close to the maximum value 1, the current running state and the surrounding environment of the host vehicle are the same as the situation where an accident or an accident is likely to occur in the past. It is possible to provide driving assistance effectively and accurately by, for example, alerting the driver or performing an action to avoid a collision. It becomes.

  Further, as in the present embodiment, past risk information data is accumulated as data 2 seconds before or 2 seconds before an event that may have caused a collision, and correlation is made with the current data. Thus, it is possible to determine whether or not the possibility of the collision is high 2 seconds before the occurrence of the event that may cause the collision. Therefore, it is possible to provide driving support with a sufficient margin. It should be noted that the number of seconds before the occurrence of an event as the past danger information is appropriately determined.

  Further, the determination means 3 determines that the driving state or the surrounding environment has a high possibility of a collision when, for example, sudden braking is performed, and accumulates each data at that time in the storage device 4. it can. Therefore, since data is accumulated as past hazard information on the vehicle, not as a general accident point on the map, it is accurate according to the driver's oversight and tendency to overlook obstacles and pedestrians. It is possible to perform driving support such as collision avoidance.

  In addition, for steering information, speed information, accelerator information, and brake information necessary as information for performing collision avoidance etc., time-series data for a certain period of time before that was used, not just data at the present time. Can provide more accurate driving support.

At this time, the direct correlation values r _st , r _v are directly extracted based on the current time-series data and the time-series data as past risk information without extracting the feature values as in the present embodiment. , R _acc , r _brk may be calculated. However, by extracting feature values and calculating their correlation as in this embodiment, it is possible to determine whether and how much the current situation is similar to the past situation where there was a possibility of a collision. It becomes possible to clarify, and the reliability and robustness of the apparatus can be improved.

  In addition, when configured to extract feature amounts as in the present embodiment, the amount of data can be reduced, and searches and calculations can be performed in a shorter time.

  In the driving support device 1 according to the present embodiment, when it can be determined from the measured value of the brake hydraulic pressure Bp that the driver is clearly taking an avoidance measure, for example, the level is lowered by one step, or an alarm is issued. It is also possible to configure so as not to utter and support driving. If comprised in this way, it can prevent that the driver | operator who is taking the collision avoidance action feels bothersome because driving assistance is act | operated contrary to will.

  In addition, in the accumulation of past risk information data in the host vehicle by the determination means 3, for example, the data when an accident actually occurs is accumulated by setting a flag or the like, and the determination means 5 Alternatively, the driving support level can be increased by one step when the surrounding environment has the highest correlation with the past data flagged.

Furthermore, the technique of the present invention can be used in combination with the conventional techniques described in Patent Documents 1 to 3, for example. In addition, it is possible to configure so that information such as ITS (Intelligent Road Traffic System) is taken into the driving support device 1. For example, such information is used instead of the information from the surrounding environment recognition device 8. It is also possible to do.
Further, general cases that may cause accidents can be stored in the storage device 4 in advance as initial information of danger information.

  Further, instead of performing feature amount extraction by a frequency spectrum analysis method using discrete Fourier transform as in this embodiment, for example, eigenvalues obtained by a principal component analysis method of a covariance matrix are extracted as feature amounts. It is also possible to configure as described above.

  In this case, for each of the steering angle δ, the vehicle speed V, the accelerator opening A, and the brake hydraulic pressure Bp, eigenvalues can be obtained by the principal component analysis method of the covariance matrix and extracted as feature amounts. From this point of view, a case where feature values are extracted collectively will be described.

First, for the steering angle δ, the vehicle speed V, the accelerator opening A, and the brake hydraulic pressure Bp, the data for 5 seconds from 7 seconds to 2 seconds before the event where there was a possibility of a collision in the past, or from 5 seconds before the present Assuming that the number of data of the data for 5 seconds until is n, respectively, n four-dimensional vectors X _k = (δ, V, A, Bp) whose components are normalized values of data for each cycle Generate.

Then, the following (25) the covariance matrix based calculates the sample mean vector μ of four-dimensional vectors X _k, said each four-dimensional vectors X _k from the sample mean vector μ below (26) on the basis of the formula R Ask for.

  Then, an eigenvalue matrix Λ and an eigenvector matrix Φ satisfying the following equation (27) are obtained by principal component analysis.

Here, the eigenvalue matrix Λ is expressed as the following equation (28) when the eigenvalues of R are λ ₁ , λ ₂ , λ ₃ , λ ₄ (λ ₁ ≧ λ ₂ ≧ λ ₃ ≧ λ ₄ ). The eigenvector matrix Φ is expressed as the following equation (29) when the eigenvectors corresponding to the eigenvalues λ ₁ , λ ₂ , λ ₃ , and λ ₄ of R are φ ₁ , φ ₂ , φ ₃ , and φ _4. The

Then, the eigenvalues λ ₁ , λ ₂ , λ ₃ , and λ ₄ obtained in this way are extracted as feature amounts.

  As described above, even when the eigenvalue obtained by the principal component analysis method of the covariance matrix is extracted as the feature quantity, the eigenvalue matrix Λ composed of the extracted eigenvalues and eigenvectors corresponding to the eigenvalues are included. It is possible to faithfully restore the original covariance matrix R from the eigenvector matrix Φ. Therefore, it can be said that the eigenvalue which is the feature quantity is characterized by the tendency of the data as in the case of using the Fourier transform, and has been degenerated so that the characteristics of the data are retained from the enormous amount of data. It can be said that it is a thing.

  In this case, the correlation values are generated as vectors having four eigenvalues obtained for the past 5 seconds of data and the current 5 seconds of data, respectively, as shown in equation (13). Obtained from the Euclidean distance between two vectors.

  Further, the eigenvector itself for the eigenvalue is also degenerated so that the characteristic principal component of the data is retained. Therefore, instead of using eigenvalues obtained by the principal component analysis method of the covariance matrix as feature quantities, it is possible to extract a certain number of eigenvectors together with eigenvalues.

  Further, in the above-described embodiment, the example in which the detection device 2, the determination unit 3, the storage device 4, the determination unit 5, and the responding unit 23 are all mounted on the own vehicle is described. It is also possible to provide the device 4 and the determination means 5 in an information management center installed outside the host vehicle. A configuration in which the storage device 4 and the determination unit 5 are provided in the information management center will be described.

  Here, the information management center performs information communication with a plurality of vehicles including the own vehicle via a wireless communication network, and centrally stores information transmitted from each vehicle as shared data used by the plurality of vehicles. In addition, the control signal for operating the response unit 23 of the specific vehicle based on the information transmitted from each vehicle is transmitted to the specific vehicle.

  The storage device 4 and the determination unit 5 provided in this information management center are connected to the determination unit 3 of the own vehicle via a communication network. The judging means 5 is also connected to the responding part 23 of the own vehicle via a communication network.

  The determination means 3 in the own vehicle is similar to the above-described embodiment, and each vector data for road shape, area information, time information, obstacle information and vehicle switch information shown in Table 1, steering information, speed information The four feature amounts max1, max2, max3, and max4 calculated for the accelerator information and the brake information are calculated as past danger information data. Then, the calculated past risk information data is transmitted to the storage device 4 provided in the information collection center.

  Further, the determination means 3 determines each data at the present time of road shape, area information, time information, obstacle information and vehicle switch information described in Table 1 input from the detection device 2 of the own vehicle via a communication network. Send to 5. At this time, each data at the present time is transmitted to the determination means 5 together with the vehicle ID for identifying the transmission source of the data.

  On the other hand, the storage unit 4 in the information management center receives and stores past risk information data transmitted from the determination unit 3. In addition, the determination means 5 in the information management center is based on the past risk information data stored in the storage device 4 and the current data transmitted from the determination means 3 in the same manner as in the above embodiment. Correlation values r for past risk information data are calculated, and the correlation values r are classified by threshold values to determine whether or not there is a possibility of a collision.

  When the determination means 5 determines that there is a possibility of a collision, the determination means 5 refers to the vehicle ID to identify the vehicle that has transmitted each data at the present time, and sends an operation signal to the response unit 23 of the identified vehicle. Send.

  The response unit 23 of the vehicle operates the response unit 23 based on the received operation signal to alert the driver or avoid collision.

  According to such a configuration, the past risk information data calculated by the determination means 3 can be used as shared data used by a plurality of vehicles, whereby the risk information data is stored only by the host vehicle. It is possible to accumulate risk information data at an early stage compared to.

It is a block diagram which shows the structure of the driving assistance apparatus which concerns on this embodiment. It is a block diagram which shows the structure of the solid-object monitoring apparatus as surrounding environment recognition apparatus. It is a figure which shows an example of the captured image imaged with the imaging device. It is a figure which shows the distance image calculated from the captured image of FIG. It is a figure which shows the distance image divided | segmented into the division at a predetermined interval in the horizontal direction. It is a figure which shows the solid object detected as an "object" or a "side wall". It is a figure showing the database structure in a memory | storage device.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Driving assistance apparatus 2 Detection apparatus 3 Judgment means 4 Storage apparatus 5 Judgment means 23 Response part r Correlation value

Claims (3)

In a driving support device that determines the possibility of a collision of the host vehicle and supports driving of the vehicle,
A detection device for detecting data relating to the running state of the host vehicle and data relating to the surrounding environment of the host vehicle ;
Determination means for determining whether or not the driving state and the surrounding environment have a high possibility of collision based on the data detected by the detection device;
When the determination unit determines that the driving state and the surrounding environment have a high possibility of a collision, the data on the driving state of the own vehicle and the data on the surrounding environment of the own vehicle at a time point that is a predetermined time later than the predetermined state the data obtained from Step a storage device for storing the data of the past danger information in the vehicle,
A correlation value indicating a correlation state between the data relating to the traveling state of the own vehicle at the time of the current sampling detected by the detection device and the data relating to the surrounding environment of the own vehicle and the data of the past danger information stored in the storage device; calculated, the driving support apparatus correlation value calculated is equal to or obtaining Bei a determination means for determining whether the possibility of collision at the time of the current sampling by whether exceeds a preset threshold.   The travel support apparatus according to claim 1, wherein when the determination unit determines that there is a possibility of a collision based on the correlation value, the predetermined response unit of the host vehicle is operated. The determination means calculates a feature amount from data detected by the detection device when it is determined that the traveling state has a high possibility of a collision, and the feature amount is used as past risk information data in the host vehicle. Storing in the storage device;
The determination unit calculates a feature amount from data detected by the detection device, and calculates a correlation value between the calculated feature amount and the feature amount as data of the past danger information stored in the storage device. The travel support apparatus according to claim 1 , wherein

Images