WO2021255982A1

WO2021255982A1 - Driving assistance device

Info

Publication number: WO2021255982A1
Application number: PCT/JP2021/004259
Authority: WO
Inventors: 寛人三苫; 健永崎
Original assignee: 日立Astemo株式会社
Priority date: 2020-06-18
Filing date: 2021-02-05
Publication date: 2021-12-23
Also published as: JP7406635B2; DE112021002185T5; JPWO2021255982A1

Abstract

Provided is a driving assistance device capable of more appropriate brake control. The present invention calculates the predicted travel path of a vehicle and the predicted travel path of an oncoming vehicle on the basis of the ground speed of the oncoming vehicle, the relative positions of the vehicle and the oncoming vehicle, and the state of the vehicle, determines the possibility of collision with the oncoming vehicle when turning right or left of the vehicle, calculates the estimated time of collision between the vehicle and the oncoming vehicle, and controls the brakes of the vehicle on the basis of the amount of movement, including at least the lateral movement of the vehicle as seen from the oncoming vehicle, during the period from the estimated collision time to a predetermined time before.

Description

Driving support device

The present invention relates to a driving support device that detects another vehicle and avoids / reduces a collision with the other vehicle.

As a driving support device for avoiding / reducing a collision with another vehicle, for example, there is one disclosed in Patent Document 1. Patent Document 1 describes that brake control is performed in consideration of a collision site with another vehicle and a collision angle.

Japanese Unexamined Patent Publication No. 2009-208560

In the above-mentioned Patent Document 1, brake control is performed from the degree of influence of the collision in consideration of the collision site and the collision angle. The same movement is expected, and a collision with an oncoming vehicle existing in an oncoming lane that does not actually collide is expected, and there is a possibility that unnecessary brake control may be performed by mistake.

The present invention has been made in consideration of the above situation, and an object of the present invention is to provide a driving support device capable of performing more appropriate brake control.

In order to solve the above problems, the driving support device according to the present invention has the predicted traveling path of the own vehicle and the said vehicle based on the ground speed of the oncoming vehicle, the relative position between the own vehicle and the oncoming vehicle, and the state of the own vehicle. The predicted travel path of the oncoming vehicle is calculated, the possibility of collision with the oncoming vehicle at the time of right turn or left turn of the own vehicle is determined, and the estimated collision time at which the own vehicle and the oncoming vehicle collide is calculated. The brake of the own vehicle is controlled based on the movement amount of the own vehicle from the expected collision time to a predetermined time before, and the movement amount is at least lateral movement of the own vehicle as seen from the oncoming vehicle. It is characterized by including an amount.

According to the present invention, the brake can be applied as needed, so that more appropriate brake control can be performed.

Issues, configurations and effects other than those described above will be clarified by the explanation of the following embodiments.

The whole block diagram of the driving support apparatus which concerns on one Embodiment of this invention. Functional block diagram of a stereo camera. An explanatory diagram of the distance measurement principle of the distance calculation processing unit. An explanatory diagram of the distance calculation processing operation of the distance calculation processing unit. An explanatory diagram of the image matching degree of the distance calculation processing unit. An explanatory diagram of the image matching degree of the distance calculation processing unit. Explanatory drawing of the collision determination processing operation of the collision determination processing unit. Explanatory drawing of the operation scene of this embodiment. Explanatory drawing of the non-operation scene of this embodiment. Explanatory drawing of the non-operation scene of this embodiment. Explanatory drawing of processing operation at the time of an oncoming vehicle curve.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 shows a diagram illustrating the overall configuration of the driving support device according to the present embodiment. The driving support device 100 includes a stereo camera 101 as an outside world recognition sensor, a vehicle information unit 102 that acquires vehicle information indicating the state of the own vehicle such as yaw rate and own vehicle speed, a brake control unit 103 as a unit that controls the vehicle, and a brake. It includes an actuator 104, these sensors, and a CAN (Controller Area Network) 105 that transmits information between the units. The driving support device 100 shown in FIG. 1 is mounted on a vehicle, estimates a collision between the own vehicle and another vehicle, and if it is estimated that the collision occurs, brake control is performed to avoid / reduce the collision. Is.

The vehicle information unit 102 transmits to the CAN 105 the yaw rate at which the change in the turning angle in the yaw direction applied to the own vehicle is detected and the own vehicle speed detected from the number of rotations of the wheels.

The stereo camera 101 is configured as shown in FIG. The

cameras

201 and 202 are installed near the rear-view mirror in the vehicle interior at the same height at a certain distance in the horizontal direction so that the front of the vehicle is photographed. These cameras are equipped with image sensors such as CCD and CMOS, are synchronized with each other, and are sampled at the same timing.

The image acquisition unit 203 and the image acquisition unit 204 convert the brightness values of the images captured by the camera 201 and the camera 202 into digital so that the subsequent processing units can process the images, and the images are captured by the camera 201 and the camera 202, respectively. The image is corrected so as to eliminate the difference in the imaging environment and imaging characteristics between the two cameras, and the image data is passed to the next process.

The distance calculation processing unit 205 divides the image acquired by the image acquisition unit 203 into fixed block sizes (for example, 4 × 4 [pix]), and calculates the distance in real space for each divided block. FIG. 3 shows a general distance measurement principle using a stereo camera. In the figure, the symbol D represents the distance from the surface of the lens 302 and the lens 303 to the measurement point 301, f represents the distance (focal length) between the lens 302 and the lens 303 and the imaging surface 304 and the imaging surface 305, and b. Represents the distance (focal length) between the center of the lens 302 and the lens 303, d is the position where the measurement point 301 is imaged on the image pickup surface 304 through the lens 302, and the measurement point 301 is imaged on the image pickup surface 305 through the lens 303. Indicates the difference (parallax) from the position of the lens. The following mathematical formula (1) holds between these symbols due to the similarity relationship of triangles.
(Equation 1) D = b × f / d ... (1)

From the mathematical formula (1), the focal length f and the baseline length b are constants determined by the configuration of the camera. Therefore, in order to obtain the distance D, the parallax d, which is the difference in appearance between the left and right lenses, may be obtained. A method for obtaining this parallax d will be described with reference to FIG. FIG. 4 shows an image captured by the camera 201 as a left image (reference image) and an image captured by the camera 202 as a right image (reference image) when the camera 201 is installed on the left and the camera 202 is installed on the right. There is. As shown in FIG. 4, an image 401 having a block size (for example, 4 × 4 [pix] size) in which the reference image is divided by a fixed size and an image having the same height and size as the image 401 on the reference image. The same target image is found from the right image by shifting from 402 to the image 403, which is a certain pixel away, one pixel at a time, calculating the degree of matching for each pixel, and searching for the block with the highest degree of matching. It is possible to calculate the parallax d of the block 401 of the left image. The degree of matching can be calculated by using the sum of the absolute values of the luminance differences between the pixels (Sum of the Absolute Differences: SAD). At this time, if the horizontal axis is the parallax [pix] and the vertical axis is the degree of agreement (SAD), the graph is as shown in FIG. 5, and the minimum value 501 is the parallax with the highest degree of agreement. From the above, since the parallax d can be obtained, the distance D in the real space for each block can be calculated by using the mathematical formula (1). Also, if there is no texture in the block size image 401 that divides the reference image by a fixed size, or if the same image cannot be found on the left and right (for example, the windshield in front of the camera is on the left and right of the same target due to the influence of raindrops etc. (When the images are different) is treated as invalid parallax because, as shown in FIG. 6, although the minimum value 601 exists, the degree of coincidence is not relatively low and the reliability is low. Whether to use effective parallax or invalid parallax is determined by whether the minimum value of the degree of coincidence is below the threshold value and the difference between the average degree of coincidence and the minimum value in the periphery exceeds the threshold value.

The three-dimensional object extraction processing unit 206 extracts a three-dimensional object using the distance image obtained by the distance for each block over the entire image obtained by the distance calculation processing unit 205. First, create a histogram divided by a fixed distance unit for each block sequence of the distance image, and if the number of histograms of the frequently peaked distance (representative distance) is equal to or greater than the threshold (variable according to the distance). , It is assumed that there are three-dimensional object candidates in the block row. Next, in the adjacent block row, when there are three-dimensional object candidates and the difference between the representative distances at the peak of the histogram is equal to or less than the threshold value (variable according to the distance), they are grouped as the same three-dimensional object. Finally, when the width of the grouped three-dimensional object is equal to or larger than the threshold value (variable according to the distance), it is registered as a three-dimensional object. After being registered as a three-dimensional object, the blocks around the distance of the row that became the same three-dimensional object are grouped on the distance image, and the position on the distance image (lower end, upper end, left end, right end on the image) and grouping. The distance to a three-dimensional object in real space is calculated by averaging the distances. Hereinafter, the three-dimensional object obtained as a result is referred to as an extracted three-dimensional object (having parameters such as the relative position, height, and width with the own vehicle).

The three-dimensional object tracking processing unit 207 tracks the extracted three-dimensional object obtained by the three-dimensional object extraction processing unit 206 in time series, and subtracts the change in the relative position due to the behavior of the own vehicle, and subtracts the change in the relative position. (Front-back speed, left-right speed) is calculated, and parameters such as relative position, height, width, and ground speed with the own vehicle are filtered, and the tracking solid object (relative to the filtered own vehicle) is performed. It has the position, height, width, ground speed, etc. as parameters). More specifically, it is realized by executing the following processing in each frame. If there is a tracking solid object of the previous processing frame, this time from the change due to the behavior of the own vehicle (the amount of position change that moves relatively from the yaw rate and the own vehicle speed) and the ground speed parameter of the tracking solid object. Calculate the predicted position that will exist in the processing frame of. This predicted position is compared with the extracted three-dimensional object detected in the current processing frame, and if the difference between each parameter is less than or equal to the threshold value, it is judged to be the same object, and the parameters of the traced three-dimensional object in the previous processing frame. Is updated using the parameters of the extracted three-dimensional object of this processing frame. The details of the update method will not be described, but the error variance can be realized by setting the extracted three-dimensional object as an observed value using a Kalman filter, for example, and obtaining the error variance from the actually measured value. Further, the extracted three-dimensional object that does not match any of the tracking three-dimensional objects of the previous processing frame is newly registered in the tracking three-dimensional object as it is as the initial detection three-dimensional object. At this time, since the speed cannot be calculated with only one frame, the ground speed is set to zero.

Based on the information of the traced three-dimensional object, the three-dimensional object tracking processing unit 207 can estimate the traveling path (predicted traveling path) of the tracking three-dimensional object.

The CAN information acquisition unit 208 acquires the yaw rate and the own vehicle speed transmitted from the vehicle information unit 102 via the CAN 105. Based on this acquired information, the own vehicle traveling path estimation processing unit 209 estimates the traveling path (predicted traveling path) of the own vehicle. Specifically, from the acquired vehicle speed V [m / s] and yaw rate r [rad / s], the turning radius R [m] of the vehicle is obtained using the following equation (2), and this is calculated. It is used as the estimated course (predicted course) of the own vehicle.
(Number 2) R = V / r ... (2)

In the collision determination processing unit 210, as shown in FIG. 7, the estimated traveling path 702 of the own vehicle 701 calculated from the turning radius R of the own vehicle 701 and the tracking three-dimensional object 703 calculated from the ground speed of the tracking three-dimensional object 703. From the estimated travel path 704, it is determined whether or not the own vehicle 701 and the tracking three-dimensional object 703 collide with each other. Specifically, using the following equations (3) and (4), the predicted front-rear direction position x_s (t) and the left-right direction position y_s (t) of the own vehicle 701 are set in 0.1 second units for 3 seconds. Minutes (t = 0.1, 0.2, 0.3, ..., 2.9, 3.0) are calculated.
(Number 3) x_s (t) = Rsin (r × t) ... (3)
(Number 4) y_s (t) = R (1-cos (r × t)) ... (4)

Further, using the following equations (5) and (6), the expected front-back direction position x_o (t) and the left-right direction position y_o (t) of the tracking three-dimensional object 703 are calculated for 3 seconds in 0.1 second units. do. However, the ground speed in the front-rear direction as seen from the current coordinate system of the own vehicle 701 is xv_t (0), and the ground speed in the left-right direction is yv_o (0).
(Equation 5) x_o (t) = x_o (0) + xv_o (0) × t ... (5)
(Number 6) y_o (t) = y_o (0) + yv_o (0) × t ... (6)

Next, using the values obtained by the equations (3), (4), (5), and (6), the distance d (t) between the own vehicle 701 and the tracking three-dimensional object 703 at each time t is set as follows. It is obtained by the formula (7).
(Number 7) d (t) = √ ((x_s (t) -x_o (t)) ^ 2
+ (Y_s (t) -y_o (t)) ^ 2) ... (7)

When the minimum value of the obtained distance d (t) is equal to or less than a predetermined threshold value (for example, 1 m), t is set as the expected collision time t_hit, and if the estimated collision time exists, the collision with the tracking three-dimensional object 703 is performed. Judge that there is a possibility.

In the lateral movement amount determination processing unit 211, when the collision determination processing unit 210 determines that there is a possibility of collision, what kind of control is performed on the own vehicle 701 using the obtained estimated collision time t_hit. Determine the amount of lateral movement to determine. Specifically, formula (3), ( 4) is used to calculate t = t_hit and t_hit-t_past, respectively. At this time, the traveling direction angle θ_o of the tracking three-dimensional object 703 as seen from the coordinate system of the own vehicle 701 is obtained by the following equation (8).
(Equation 8) θ_o = π / 2-atan (xv_o (0) / yv_o (0)) ... (8)

Using the following equation (9) using this angle, the lateral position y_so (t_hit-t_past) of the own vehicle 701 as seen from the tracking three-dimensional object 703 before the predetermined time t_past from the expected collision time t_hit is calculated, and the following equation (t_hit-t_past) is calculated. 10), the absolute value d_h [m] is calculated as the amount of change in the lateral position (also referred to as the amount of lateral movement) 707.
(Equation 9) y_so (t) = y_s (t) cos (θ_o) -x_s (t) sin (θ_o) ... (9)
(Number 10) d_h = | y_so (t_hit-t_past) | ... (10)

That is, the lateral position change amount d_h obtained here is the lateral position change amount (lateral movement amount) of the own vehicle 701 as seen from the tracking three-dimensional object 703 from the expected collision time t_hit to before the predetermined time t_past. It corresponds to the length of the vertical line drawn from the position 706 of the own vehicle 701 before the predetermined time t_past from the estimated collision time t_hit to the estimated traveling path 704 (here, assuming straight running) of the tracking three-dimensional object 703.

The control processing unit 212 determines what kind of control is to be performed on the own vehicle 701 by using the lateral position change amount d_h obtained by the lateral movement amount determination processing unit 211. When the obtained lateral position change amount d_h is larger than a predetermined threshold value (for example, 5 m), a control command for operating the brake actuator 104 is given to the brake control unit 103 as a unit that controls the vehicle without performing brake control (that is,). (Not transmitted), if it is below a predetermined threshold (only), brake control is performed (that is, a control command is transmitted to the brake control unit 103 as a unit that controls the vehicle to operate the brake actuator 104).

As a result, at the intersection of one lane on each side as shown in FIG. 8, the own vehicle 801 turns right on the track of the estimated travel path 802 of the own vehicle 801 and the oncoming vehicle 803, which is a tracking three-dimensional object, turns to the estimated travel path 804 of the oncoming vehicle 803. In the scene of going straight on the track, the lateral position change amount 807 (d_h) is about one lane width (about 3.5 m), and the lateral position change amount d_h is equal to or less than a predetermined threshold, so it is necessary. Brake control can be performed.

On the contrary, when entering the right turn lane by changing lanes before the intersection as shown in FIG. 9, the movement is temporarily the same as that of the right turn as shown in the estimated travel path 902 of the own vehicle 901, and the vehicle actually collides. It is expected that the vehicle will collide with the oncoming vehicle 903 that goes straight on the track of the estimated traveling path 904 that exists in the oncoming lane, but the lateral position change amount 907 (d_h) is a large value of about two lane widths (about 7 m). Therefore, the lateral position change amount d_h is larger than a predetermined threshold value, and erroneous braking control can be prevented.

Further, a scene in which the own vehicle 1001 shown in FIG. 10 turns right on the track of the estimated traveling path 1002 of the own vehicle 1001 and the oncoming vehicle 1003 goes straight on the estimated traveling path 1004 of the oncoming vehicle 1003 (the oncoming lane is two lanes and the own vehicle). In the scene where the oncoming vehicle 1003 travels in the lane away from the lane of 1001), the lateral position change amount 1007 (d_h) is as large as about two lane widths (about 7 m), and the lateral position change amount d_h. Is larger than a predetermined threshold value, so that the brake control is not performed even in this scene, but when the lateral position change amount 1007 (d_h) seen from the oncoming vehicle 1003 is large, the own vehicle 1001 moves from the oncoming vehicle 1003. It is easy to see if it is, and there is a high possibility that the oncoming vehicle 1003 will also apply the brakes, so it is considered that there is no need to apply the brakes by the system. Further, when the lateral position change amount 1007 (d_h) seen from the oncoming vehicle 1003 is small, it is difficult to visually recognize whether or not the own vehicle 1001 is moving, and the possibility that the oncoming vehicle 1003 will brake is low. It seems necessary to apply the brakes. In the present embodiment described above, it is possible to perform appropriate brake control even in this scene.

On the other hand, it is possible to detect the white line and predict the subsequent movement of the own vehicle, but in the actual environment, the white line may not be visible due to snow or rainfall, so this embodiment does not use the white line detection. Is considered valid.

In addition, by using the amount of lateral position change of the own vehicle as seen from the oncoming vehicle from the expected collision time to the time before the predetermined time, for example, the vehicle is stopped once in the right turn lane while the vehicle is tilted to the right. Since the amount of lateral movement is calculated to be smaller (than the amount of lateral movement seen from the own vehicle) even when the vehicle starts to move from, the brake can be operated appropriately even in a scene where it is difficult to see from the oncoming vehicle. can.

The threshold value for the lateral position change amount (lateral movement amount) d_h used for determining whether or not to perform brake control can be arbitrarily set, but considering the scenes shown in FIGS. 8 to 10, the lane is taken into consideration. It is considered desirable that the width is one or more and the lane width is less than two. Alternatively, when the oncoming lane is two or more lanes, it is one or more lane widths of the lane closest to the own vehicle (on the side of the own vehicle's driving lane) among the oncoming lanes, and is closer to the own vehicle among the oncoming lanes. It is considered desirable that the width of the lane (on the side of the driving lane of the own vehicle) is less than two lanes.

As described above, the driving support device 100 of the present embodiment has the predicted traveling path of the own vehicle and the said vehicle based on the ground speed of the oncoming vehicle, the relative position between the own vehicle and the oncoming vehicle, and the state of the own vehicle. The predicted travel path of the oncoming vehicle is calculated, the possibility of collision with the oncoming vehicle at the time of right turn or left turn of the own vehicle is determined, and the estimated collision time at which the own vehicle and the oncoming vehicle collide is calculated. The brake of the own vehicle is controlled based on the movement amount of the own vehicle from the expected collision time to a predetermined time before, and the movement amount is at least lateral movement of the own vehicle as seen from the oncoming vehicle. Includes quantity.

Further, the brake of the own vehicle is activated only when the lateral movement amount of the own vehicle is equal to or less than a predetermined threshold value, and when the lateral movement amount of the own vehicle is larger than the predetermined threshold value, the brake of the own vehicle is activated. Does not work.

According to the present embodiment described above, the brake can be applied as needed, so that the brake control can be performed at a more appropriate timing.

In the above embodiment, the predicted course of the oncoming vehicle is assumed to be a straight line, but the turning radius of the oncoming vehicle may be estimated and used as the predicted course of the curve. Specifically, as shown in FIG. 11, while the own vehicle 1101 is turning right on the track of the predicted course 1101 of the own vehicle 1101, the oncoming vehicle 1103 curves to the right on the track of the predicted course 1104 of the oncoming vehicle 1103. If the above is the case, the side of the oncoming vehicle 1103 in the normal direction with respect to the predicted traveling path 1104 at the position 1106 of the own vehicle 1101 a predetermined time before the expected collision position 1105 (the position 1105 of the own vehicle 1101 at the expected collision time t_hit). The position is calculated, and the absolute value is used as the lateral position change amount (lateral movement amount) 1107. That is, the length of the normal line from the position 1106 of the own vehicle 1101 one predetermined time before the predicted collision time t_hit to the predicted traveling path 1104 (here, assuming curve driving) of the oncoming vehicle 1103 is viewed from the oncoming vehicle 1103. The lateral position change amount (lateral movement amount) of the own vehicle 1101 is set to 1107. As a result, an appropriate amount of lateral position change can be calculated even in a scene in a curve, and a more appropriate determination can be made.

Further, in the above embodiment, the predicted traveling path of the own vehicle is simply calculated from the turning radius based on the current parameter, but in order to calculate the amount of lateral position change seen from the oncoming vehicle more accurately, the actual past actual amount is calculated. You may keep the course for a certain period of time and use it to calculate the predicted course of your vehicle.

In addition, the predicted course of the oncoming vehicle was calculated using a stereo camera (image acquired by) having two cameras, but the sensor information acquired by an external recognition sensor (acquired by) such as a monocular camera, a millimeter wave radar, and a laser radar. ) May be used to calculate the predicted course of the oncoming vehicle. Further, the predicted traveling path of the oncoming vehicle may be calculated by acquiring the oncoming vehicle information (ground speed, position, predicted traveling path, etc.) via communication with an external server or inter-vehicle communication.

Further, in the above embodiment, the presence / absence of the brake control is switched according to the lateral position change amount, but the brake control timing may be changed according to the lateral position change amount. For example, when the lateral position change amount is larger than a predetermined threshold value, the brake control timing (that is, the brake operation timing) may be delayed as compared with the case where the lateral position change amount is equal to or less than the predetermined threshold value.

The above embodiment is assumed to avoid / reduce a collision with an oncoming vehicle when making a right turn at an intersection, but it is of course applicable to the same when making a left turn.

The present invention is not limited to the above-described embodiment, but includes various modified forms. For example, the above-described embodiment has been described in detail in order to explain the present invention in an easy-to-understand manner, and is not necessarily limited to the one including all the described configurations.

Further, each of the above configurations, functions, processing units, processing means, etc. may be realized by hardware by designing a part or all of them by, for example, an integrated circuit. Further, each of the above configurations, functions, and the like may be realized by software by the processor interpreting and executing a program that realizes each function. Information such as programs, tables, and files that realize each function can be stored in a memory, a hard disk, a storage device such as an SSD (Solid State Drive), or a recording medium such as an IC card, an SD card, or a DVD.

In addition, the control lines and information lines indicate those that are considered necessary for explanation, and do not necessarily indicate all control lines and information lines in the product. In practice, it can be considered that almost all configurations are interconnected.

100 Driving support device 101 Stereo camera 102 Vehicle information unit 103 Brake control unit 104 Brake actuator 105 CAN
201, 202

Camera

203, 204 Image acquisition unit 205 Distance calculation processing unit 206 Three-dimensional object extraction processing unit 207 Three-dimensional object tracking processing unit 208 CAN information acquisition unit 209 Vehicle travel path estimation processing unit 210 Collision detection processing unit 211 Lateral movement amount determination Processing unit 212 Control processing unit

Claims

Based on the ground speed of the oncoming vehicle, the relative position between the own vehicle and the oncoming vehicle, and the state of the own vehicle, the predicted traveling path of the own vehicle and the predicted traveling path of the oncoming vehicle are calculated.
The possibility of collision with the oncoming vehicle when the own vehicle makes a right turn or a left turn is determined, and the possibility of collision is determined.
Calculate the expected collision time when the own vehicle and the oncoming vehicle collide with each other.
A driving support device that controls the brake of the own vehicle based on the amount of movement of the own vehicle from the expected collision time to a predetermined time before.
The driving support device, characterized in that the movement amount includes at least the lateral movement amount of the own vehicle as seen from the oncoming vehicle.
The lateral movement amount of the own vehicle as seen from the oncoming vehicle is the length of a vertical line or a normal line from the position of the own vehicle to the predicted traveling path of the oncoming vehicle from the position of the own vehicle before the predetermined time from the predicted collision time. The driving support device according to claim 1, characterized in that it is present.
The brake of the own vehicle is activated only when the lateral movement amount of the own vehicle is equal to or less than a predetermined threshold value, and when the lateral movement amount of the own vehicle is larger than the predetermined threshold value, the brake of the own vehicle is activated. The driving support device according to claim 1, wherein the driving support device is not allowed to be used.
The driving support device according to claim 3, wherein the predetermined threshold value is one lane width or more and less than two lane widths.
The driving support device according to claim 1, wherein the operation timing of the brake of the own vehicle is changed according to the lateral movement amount of the own vehicle.