JP2019128153A - Distance calculation device and vehicle control device - Google Patents

Abstract

To provide a distance calculation device and a vehicle control processing which, in calculating a distance between a vehicle and an object, enables a region ensuring excellent calculation accuracy to be enlarged, thereby improving practicality.SOLUTION: A distance calculation device 1 acquires an actual measurement distance Xs_o_n to an inside of an overlap region 7c using image data from left and right cameras 4a and 4b and calculates first and second estimation distances Xn_o_n and Xn_o_m to an inside of the overlap region 7c and an inside of a left nonoverlap region 7a, respectively, using the image data from the left camera 4a. In addition, the distance calculation device calculates such a correction value Kmod as decreases an absolute value of a difference E_id_n between the actual measurement distance Xs_o_n and the first estimation distance Xn_o_n when an object is in the overlap region 7c and calculates a calculation distance X_h by correcting an estimation distance Index_x_h with the correction value Kmod when the object is in the two regions 7a and 7c.SELECTED DRAWING: Figure 2

Description

  The present invention relates to a distance calculation device and a vehicle control device that are mounted on a vehicle and calculate a distance between the vehicle and an object.

  Conventionally, what was described in patent document 1 as a distance calculation apparatus is known. This distance calculation device is of a stereo camera type mounted on a vehicle and includes a pair of left and right cameras. In the pair of left and right cameras, the optical axes of the two are diagonally disposed so that the distance between the two becomes wider toward the front, and the respective imaging regions do not overlap with the stereoscopic viewing regions overlapping each other. It consists of a viewing area.

  In the case of the distance calculation device according to the second embodiment, when there is no pedestrian in front of the host vehicle, the distance between the host vehicle and the preceding vehicle is calculated based on the left and right captured images in the stereo viewing area by the stereo matching method. It is calculated.

JP 2005-24463 A

  According to the above-described conventional distance calculation device, since the distance between the subject vehicle and the object can be calculated only under the condition that the object such as the leading vehicle is located in the stereo vision region, the object is In the case where it exists across the stereo area and the non-stereo view area, there is a problem that the distance of the object in the non-stereo area cannot be calculated with high accuracy and the practicality is low.

  The present invention has been made to solve the above-mentioned problems, and in the case of calculating the distance between a vehicle and an object, it is possible to expand an area where good calculation accuracy can be secured, and to improve practicability. An object of the present invention is to provide a distance calculation device and a vehicle control process that can perform the above.

  In order to achieve the above object, the present invention is mounted on a vehicle 3 and in a distance calculation device 1 for calculating a distance between the vehicle 3 and an object, distance information acquisition means (cameras 4a, Measurement distance acquisition means (ECU 2) for acquiring an actual measurement distance Xs_o_n, which is an actual measurement value of the distance between the radar 4b and the radar 8) and any point in the first predetermined area (the overlapping areas 7c and 7e) using distance information. , Measured distance calculation unit 11), image information acquisition means (left camera 4a, camera 4c) for acquiring image information, and image information and a predetermined estimation algorithm (deep learning method). An estimated distance (first and second) which is an estimated value of a distance to any point in a second predetermined area (areas 7a + 7c, 7d + 7d + 7e) wider than the first predetermined area including all the areas 7c and 7e). (2) Estimated distance calculation means (ECU 2, estimated position calculation unit 12) for calculating the estimated distance Xn_o_n, Xn_o_m), and at least a part of the object acquired by the measured distance acquisition means when positioned within the first predetermined area The difference (error) between the measured distance Xs_o_n of the object and the estimated distance (first estimated distance Xn_o_n) of the object calculated by the estimated distance calculation means when at least a part of the object is located within the first predetermined area Correction value calculation means (ECU2, correction value calculation unit 13, correction position calculation unit 14) for calculating a correction value Kmod for correcting the estimated distance of the object so that the absolute value of the signal value E_id_n) is reduced; The estimated distance Index_x_h of the object calculated by the estimated distance calculating means when the object is located in the second predetermined area including the first predetermined area By correcting the correction value Kmod, corrected distance calculating means (ECU 2, the correction position calculating section 14) for calculating a corrected distance (calculated distance X_h) of the object and is preferably provided with a.

  According to this distance calculation device, distance information is acquired by the distance information acquisition means, and using this distance information, an actual measurement distance that is an actual measurement value of a distance to any point in the first predetermined area is acquired, Image information is acquired by the image information acquisition means, and using the image information and a predetermined estimation algorithm, any point in a second predetermined area wider than the first predetermined area, including all the first predetermined areas An estimated distance that is an estimated value of the distance to is calculated. In this case, since the estimated distance is the distance to any point in the second predetermined area wider than the first predetermined area including all the first predetermined areas, it is calculated as the distance in the area wider than the actual measurement distance. It will be done.

  Furthermore, when the measured distance of the target object acquired by the measured distance acquisition means when at least a part of the target object is located within the first predetermined area, and when at least a part of the target object is positioned within the first predetermined area The correction value for correcting the estimated distance of the object is calculated so that the absolute value of the difference between the estimated distance and the estimated distance of the object calculated by the estimated distance calculation means decreases. A function is provided to reduce the error between the measured distance of the object and the estimated distance of the object when at least a part of the object is located within one predetermined area.

  Therefore, after the correction of the object by correcting the estimated distance of the object calculated by the estimated distance calculation means when the object is positioned within the second predetermined area including the inside of the first predetermined area Since the distance is calculated, the corrected distance can be calculated as a value obtained by improving the calculation accuracy of the estimated distance of the object located in the second predetermined area. As a result, in the case of calculating the distance between the vehicle and the object using the distance information and the image information, it is possible to expand the area in which good calculation accuracy can be ensured, and to improve practicability ((1) Note that “acquisition” such as “acquire actual distance” in this specification is not limited to directly detecting this value by a sensor or the like, but includes calculating / estimating this value based on a parameter).

  In the present invention, the distance information acquisition unit acquires distance information by the stereo matching method using the first camera (left camera 4a) and the second camera (right camera 4b) having higher sensitivity than the first camera. Preferably, the image information acquisition means acquires image information using the first camera (left camera 4a).

  According to this distance calculation device, the distance information is acquired by the stereo matching method using the first camera and the second camera having higher sensitivity than the first camera, and the actual distance is acquired using the distance information. Therefore, the measured distance can be obtained with high accuracy. Further, image information is acquired using the first camera, and the estimated distance is calculated using the image information and the estimation algorithm. Therefore, the estimated distance can be calculated while securing a wide viewing area. Then, as described above, since the corrected distance is calculated as a value obtained by improving the calculation accuracy of the estimated distance of the object located in the second predetermined area, the second predetermined area is wider than the first predetermined area. The corrected distance can be calculated while securing high calculation accuracy.

  In the present invention, it is preferable that the distance information acquisition means acquire distance information using one of the radar 8 and the LIDAR, and the image information acquisition means acquire image information using the camera 4c.

  According to this distance calculation apparatus, since distance information is acquired using one of the radar and LIDAR, the distance information can be obtained even under conditions where the acquisition accuracy of distance information by a camera image such as rain or fog is lowered. Can be acquired with high accuracy, and the acquisition accuracy of the measured distance can be improved. Moreover, since image information is acquired using a camera, an estimated distance can be calculated while ensuring a wide visual field area. As a result, as described above, since the corrected distance is calculated as a value obtained by improving the calculation accuracy of the estimated distance of the object located in the second predetermined area, the second predetermined area is wider than the first predetermined area. Within the region, the corrected distance can be calculated while securing high calculation accuracy.

  In the present invention, the correction value Kmod is a first predetermined section with respect to the horizontal coordinates in the two-dimensional coordinate system in which the horizontal direction relative to the reference point of the vehicle 3 is one coordinate and the vertical direction relative to the reference point is one coordinate. It is preferable to calculate as a function value which is continuous at distances Index_x_1 to Index_x_N + M and is continuous at the second predetermined section (estimated lateral position Index_y_1 to Index_y_N + M) with respect to the coordinate in the vertical direction.

  According to this distance calculation apparatus, the correction value is a first predetermined section with respect to the horizontal coordinate in a two-dimensional coordinate system in which the horizontal direction with respect to the reference point of the vehicle is one coordinate and the vertical direction with respect to the reference point is one coordinate. Because it is calculated as a function value that is continuous at the same time in the second predetermined section with respect to the coordinate in the vertical direction at the same time, when the object is located in the area on the front side of the vehicle The corrected distance can be calculated while compensating for the influence. Thereby, the calculation accuracy of the corrected distance can be further improved.

  The vehicle control device according to the present invention includes any one of the above-described distance calculation devices 1, and the first predetermined region and the second predetermined region are one of the front and rear regions of the vehicle 3, and the correction of the object It is preferable to control at least one of the driving force, braking force, and steering amount of the vehicle 3 using the rear distance (calculated distance X_h) (FIG. 7 / STEPs 1 to 3).

  According to this vehicle control device, at least one of the driving force, the braking force, and the steering amount of the vehicle can be controlled using the corrected distance with high calculation accuracy as described above, so the control accuracy is improved. It can be done.

BRIEF DESCRIPTION OF THE DRAWINGS It is a figure which shows typically the structure of the distance calculation apparatus which concerns on one Embodiment of this invention, a vehicle control apparatus, and the vehicle to which these are applied. It is a figure which shows the visual field area | region of a camera on either side. It is a block diagram which shows the functional structure of a distance calculation apparatus. It is a figure which shows an example of the map used for calculation of distance weighting function value Wx_i. It is a figure which shows an example of the map used for calculation of horizontal position weighting function value Wy_j. It is a figure which shows an example of the relationship between correction value Kmod, presumed distance Index_x, and presumed horizontal position Index_y. It is a flowchart which shows an automatic stop control process. It is a figure which shows the modification of a distance calculation apparatus.

  Hereinafter, a distance calculation device and a vehicle control device according to an embodiment of the present invention will be described with reference to the drawings. In addition, since the vehicle control apparatus of this embodiment is combined with the distance calculation apparatus, in the following description, the vehicle control apparatus will be described, and the function and configuration of the distance calculation apparatus will also be described therein.

  As shown in FIG. 1, this vehicle control device 1 is applied to a four-wheeled vehicle (hereinafter referred to as "own vehicle") 3 capable of autonomous driving, and this own vehicle 3 corresponds to left-hand traffic on a roadway. Is configured to do. The vehicle control device 1 includes an ECU 2, and left and right cameras 4 a and 4 b (see FIG. 2), a situation detection device 4 c, a prime mover 5, and an actuator 6 are electrically connected to the ECU 2.

  As shown in FIG. 2, the left and right cameras 4a and 4b have an overlapping area 7c in which the distance between the left and right cameras 4a and 4b becomes wider as the optical axes of the left and right cameras 4 move forward of the host vehicle 3 In the figure, the regions (indicated by hatching) and the non-overlapping non-overlapping regions 7a and 7b that do not overlap with each other are arranged. In this case, the angle between the optical axes of the left and right cameras 4a and 4b is set to a predetermined angle (for example, several tens of degrees) that can recognize the stop line position and the like.

  In the present embodiment, the left camera corresponds to the distance information acquisition unit, the image information acquisition unit, and the first camera, and the right camera corresponds to the distance information acquisition unit and the second camera. Further, the overlapping area 7c corresponds to the first predetermined area, and the combined area of the left non-overlapping area 7a and the overlapping area 7c corresponds to the second predetermined area.

  The left camera 4a is configured by an RGB camera that can capture images of the three primary colors, and the right camera 4b is configured by a CCC camera that can capture only monochrome images with higher sensitivity than the left camera 4a. This CCC camera outputs only clear (white) as a pixel value.

  As will be described later, the ECU 2 calculates the distance to the object located in front of the host vehicle 3 based on the image data from the left and right cameras 4a and 4b.

  Further, the situation detection device 4c is composed of a millimeter wave radar, GPS, various sensors, and the like, and outputs to the ECU 2 surrounding situation data representing the situation of the position of the vehicle 3 and the surrounding situation in the traveling direction of the vehicle 3.

  The prime mover 5 is composed of, for example, an electric motor, and the output of the prime mover 5 is controlled by the ECU 2 during execution of automatic operation control including automatic stop control described later.

  The actuator 6 includes a braking actuator, a steering actuator, and the like, and the operation of the actuator 6 is controlled by the ECU 2 during execution of automatic driving control including automatic stop control described later.

  On the other hand, the ECU 2 is constituted by a microcomputer including a CPU, a RAM, a ROM, an E2PROM, an I / O interface, various electric circuits (all not shown) and the like, and from the above-mentioned left and right cameras 4a and 4b Based on the image data and the surrounding condition data from the condition detecting device 4c, various control processes such as automatic driving control including automatic stop control described later are executed.

  In the present embodiment, the ECU 2 corresponds to measured distance acquiring means, estimated distance calculating means, correction value calculating means and corrected distance calculating means.

  Next, a functional configuration as a distance calculation device in the vehicle control device 1 of the present embodiment will be described with reference to FIG. The vehicle control apparatus 1 calculates a distance from an object located in front of the host vehicle 3 by a calculation algorithm described below, and all calculated values described below are stored in the E2PROM. .

  In this case, pedestrians and vehicles, traffic participants such as vehicles, center line, runway boundary, stop line, entrance position of intersection, entrance of entrance road after intersection, entrance of runway to turn left of intersection, center position of intersection Road structures such as traffic lights, signs, and pedestrian crossings correspond to objects.

  In the following description, the center position of the front end of the host vehicle 3 is defined as the origin (reference point), the front-rear direction of the host vehicle 3 is defined as the x coordinate axis, and the left and right direction is defined as the y coordinate axis. The coordinate value is referred to as “distance”, and the y coordinate value is referred to as “lateral position”.

  As shown in FIG. 3, the vehicle control device 1 includes an actual measurement distance calculation unit 11, an estimated position calculation unit 12, a correction value calculation unit 13, and a correction position calculation unit 14. These elements 11 to 14 are specifically Specifically, it is configured by the ECU 2.

  In this actual measurement distance calculation unit 11 (actual measurement distance acquisition means), based on the image data from the left and right cameras 4a and 4b, as the actual measurement value of the distance of the object located in the overlapping area 7c described above by stereo matching method. , N (N is a plurality) actual measurement distances Xs_o_n (n = 1 to N) are calculated. Since this stereo matching method is known, its description is omitted here.

  In the estimated position calculation unit 12 (estimated distance calculation means), N first estimated positions (Xn_o_n, Yn_o_n) and a depth learning method using a depth neural network based on the image data from the left camera 4a and M (M is a plurality) second estimated positions (Xn_o_m, Yn_o_m) are calculated.

  These N first estimated positions (Xn_o_n, Yn_o_n) are estimated values of the position of the object present in the overlapping area 7c, and N first estimated distances Xn_o_n (n = 1 to N) Coordinate values are used, and N first estimated lateral positions Yn_o_n (n = 1 to N) are used as y coordinate values.

  Also, M second estimated positions (Xn_o_m, Yn_o_m) are estimated values of the position of an object present in the left non-overlapping area 7a, and M second estimated distances Xn_o_m (m = 1 to M) Is an x-coordinate value, and M second estimated lateral positions Y n — o — m (m = 1 to M) are y-coordinate values.

  In this estimated position calculation unit 12, the position of the object on the image, the relative position between the object and another object, the movement speed (or optical flow) of the object, the object, and the like by a known deep learning method N first estimated positions (Xn_o_n, Yn_o_n) and M second estimated positions (Xn_o_m, Yn_o_m) are determined based on the speed difference between the object and the other object, the brightness / tone of the object, etc. It is calculated.

  Further, in the correction value calculation unit 13 (correction value calculation means) described above, four local correction values Kmod_ij (i = 1 to 2, j = 1 to 2) are calculated using the calculation algorithms shown in the following equations (1) to (3). 2) is calculated. In the following description, each discrete data with a symbol (k) indicates data that is sampled or calculated in synchronization with a predetermined control period ΔT (several hundred msec in the present embodiment), The symbol k (k is a positive integer) represents the control time. In the following description, the symbol (k) in each discrete data is appropriately omitted.

  E_id_n (n = 1 to N) in the above equation (1) is N error signal values, and is calculated as a deviation between the first estimated distance Xn_o_n and the actually measured distance Xs_o_n. In this case, since the actual measurement distance Xs_o_n is calculated by the stereo matching method based on the image data from the left and right cameras 4a and 4b, the first estimation is calculated based only on the image data from the left camera 4a. It can be regarded as having a calculation accuracy higher than the distance Xn_o_n. Therefore, the error signal value E_id_n can be regarded as representing an error of the first estimated distance Xn_o_n with respect to the actually measured distance Xs_o_n.

  Further, E_id_ij_n in the above equation (2) is a local error signal value. Furthermore, Wx_i (Xn_o_n) in Equation (2) is N distance weighting function values, and is calculated by searching the map shown in FIG. 4 according to the N first estimated distances Xn_o_n. On the other hand, Wy_j (Yn_o_n) in equation (2) is N lateral position weighting function values, and is calculated by searching the map shown in FIG. 5 according to N first estimated lateral positions Yn_o_n. . Furthermore, kid in the above equation (3) is a predetermined update gain that is set so that kid <0.

  As apparent from the above equations (1) to (3), the local error signal value E_id_ij_n is obtained by converting the error signal value E_id_n, which is an error of the first estimated distance Xn_o_n with respect to the actually measured distance Xs_o_n, to Xn_o_1 to Xn_o_N in the x coordinate. In the area defined by the range of Y and the range of Yn_o_1 to Yn_o_N in the y coordinate, it is calculated as a value distributed while being weighted by the four weighting function values Wx_i and Wy_j.

  Further, since the four local correction values Kmod_ij are calculated by the feedback control algorithm of only the integration term so that the absolute value of the sum of the error signal values E_id_ij_n converges to the value 0, the number of calculation times of the four local correction values Kmod_ij Is calculated so as to converge the absolute value of the error of the first estimated distance Xn_o_n with respect to the measured distance Xs_o_n to the value 0 in the above-described region as the value of L increases (that is, as updating of the four local correction values Kmod_ij proceeds). It will be

  Next, in the correction position calculation unit 14 (correction value calculation means, corrected distance calculation means), first, the correction value Kmod is calculated by the following equation (4).

  Index_x_h (h = 1 to N + M) in the above equation (4) is N + M estimated distances corresponding to a set of N first estimated distances Xn_o_n and M second estimated distances Xn_o_m. In addition, when the N first estimated distances Xn_o_n are used as the estimated distance Index_x_h, the weighting function Wx_i (Index_x_h) of the equation (4) is calculated using the map of FIG. When the second estimated distance Xn_o_m is used as the estimated distance Index_x_h, calculation is performed using a map in which the horizontal axis of FIG. 4 described above is replaced with the first estimated distance Xn_o_n to the second estimated distance Xn_o_m.

  Furthermore, Index_y_h in the above equation (4) is N + M estimated lateral positions corresponding to a set of N first estimated lateral positions Yn_o_n and M second estimated lateral positions Yn_o_m. In addition, the weighting function Wy_i (Index_y_h) of Expression (4) is calculated using the map of FIG. 5 described above when the N first estimated lateral positions Yn_o_n are used as the estimated lateral position Index_y_h, while M When this second estimated lateral position Yn_o_m is used as the estimated lateral position Index_y_h, calculation is performed using a map in which the horizontal axis of FIG. 5 described above is replaced from the first estimated lateral position Yn_o_n to the second estimated lateral position Yn_o_m. .

  Since the correction value Kmod is calculated by the above equation (4), the correction value Kmod is an absolute value of the error of the first estimated distance Xn_o_n with respect to the measured distance Xs_o_n by the function of the four local correction values Kmod_ij described above. It is calculated so as to have a function of converging.

  Also, for the same reason, in the two-dimensional coordinate system in which the estimated distance Index_x is one coordinate and the estimated lateral position Index_y is one coordinate, the correction value Kmod is continuous in a first predetermined section defined by N + M estimated distances Index_x_h. At the same time, it is calculated as a continuous function value in a second predetermined section defined by N + M estimated lateral positions Index_y_h. As a result, when the correction value Kmod is set to a coordinate axis orthogonal to the two-dimensional coordinate system, the correction value Kmod is calculated as a value that forms a correction surface (surface indicated by stippling) (see FIG. 6).

  Then, finally, N + M calculated distances X_h (h = 1 to N + M) and N + M calculated lateral positions Y_h are calculated by the following equations (5) to (6). That is, N + M calculated positions (X_h, Y_h) of the object are calculated. In the present embodiment, the calculated distance X_h corresponds to the corrected distance.

  It should be noted that the calculation of the four local correction values Kmod_ij by the above equations (1) to (3) and the calculation of the correction value Kmod by the equation (4) can be performed when stereovision at the left and right cameras 4a and 4b is performed. In the above, it is executed under the condition that an appropriate object exists across the overlapping area 4c and the left non-overlapping area 7a.

  On the other hand, when the above condition is not satisfied, for example, when stereo vision can not be performed (during tunneling, when the right camera 4b can not recognize an object such as shade and night), or an appropriate object overlaps the overlapping area 4c. And, when there is no straddling in the left non-overlap area 7a, the calculation of the four local correction values Kmod_ij and the calculation of the correction value Kmod in the correction position calculation unit 14 are stopped, and these values are within the E2PROM. Stored in the stored value.

  Next, an automatic stop control process by the vehicle control device 1 of the present embodiment will be described with reference to FIG. This automatic stop control process is executed when the vehicle 3 is under conditions to stop, and in the following description, an example will be described in which the signal of the traveling direction is red at the time of entering an intersection. .

  As shown in the figure, first, the calculated position (X_h, Y_h) of the stop line stored in the E2PROM is read (FIG. 7 / STEP1). These calculation positions (X_h, Y_h) are calculated by the calculation algorithm of the equations (1) to (6) described above in a calculation process not shown.

  Next, the prime mover 5 is controlled so that the own vehicle 3 stops appropriately at the calculated position (X_h, Y_h) of the stop line, particularly at the calculated distance X_h. Thereby, the output of the prime mover 5 is controlled (FIG. 7 / STEP2).

  Furthermore, the actuator 6 is controlled so that the host vehicle 3 appropriately stops at the calculated position (X_h, Y_h) of the stop line, particularly at the calculated distance X_h. Thereby, the braking force and the steering amount are controlled (FIG. 7 / STEP 3). Thereafter, this process is terminated.

  As described above, according to the vehicle control device 1 of the present embodiment, based on the image data from the left and right cameras 4a and 4b, the measured distance Xs_o_n of the object located in the overlapping area 7c is the stereo matching method. The first estimated position (Xn_o_n, Yn_o_n) of the object located in the overlapping area 7c and the left non-overlapping area according to the deep layer learning method using the deep layer neural network calculated and based on the image data from the left camera 4a The second estimated position (Xn_o_m, Yn_o_m) of the object located within 7a is calculated.

  Furthermore, according to the above-described equations (1) to (3), the four local correction values Kmod_ij define the error signal value E_id_n in the range of Xn_o_1 to Xn_o_N in the x coordinate and the range of Yn_o_1 to Yn_o_N in the y coordinate. The region is calculated as a value distributed while being weighted by four weighting function values Wx_i and Wy_j, and the correction value Kmod is calculated by the above-mentioned equation (4) using these four local correction values Kmod_ij. .

  In this case, the four local correction values Kmod_ij are calculated as values having the function of causing the absolute value of the error of the first estimated distance Xn_o_n to the measured distance Xs_o_n to converge to the value 0 in the above-described region. Is also calculated as a value having the same function. Therefore, N + M calculated distances X_h are calculated by correcting N + M estimated distances Index_x_h with such a correction value Kmod, and these calculated distances X_h are determined by the object located in the overlapping area 7c. The calculation accuracy of the first estimated distance Xn_o_n and the second estimated distance Xn_o_m of the object located in the left non-overlapping area 7a can be calculated as improved values. Thus, since the calculation accuracy of the calculation distance of the object located in the left non-overlapping region 7a can be improved in addition to the overlapping region 7c, the calculation accuracy better than the method of Patent Document 1. The securable area can be expanded, and practicability can be improved.

  Further, the correction value Kmod is N + M simultaneously with the first predetermined section defined by N + M estimated distances Index_x_h in a two-dimensional coordinate system with the estimated distance Index_x as one coordinate and the estimated lateral position Index_y as one coordinate. Since it is calculated as a function value that continues in the second predetermined section defined by the estimated lateral position Index_y_h, when the correction value Kmod is set to a coordinate axis orthogonal to the two-dimensional coordinate system, as shown in FIG. In addition, the correction value Kmod is calculated as a value that forms a correction surface (surface indicated by stippling). As a result, when the object is located in the area on the front side of the host vehicle 3, the calculated distance X_h can be calculated while compensating for the influence of the inclination of the host vehicle 3 in the front-rear direction and the left-right direction. Thereby, the calculation accuracy of the calculation distance X_h can be further improved.

  Further, as described above, since the automatic stop control process is executed using the calculated distance X_h with high calculation accuracy, the own vehicle 3 can be accurately stopped at the position of the stop line, and the control accuracy is improved. Can.

  Although the embodiment is an example in which N + M calculated positions (X_h, Y_h) of the object are calculated using the left and right cameras 4a and 4b, instead of this, as shown in FIG. 4d may be used in combination with the radar 8 that can detect the distance with higher accuracy. In this case, the camera 4d corresponds to an image information acquisition unit, and the radar 8 corresponds to a distance information acquisition unit.

  The camera 4 d is an RGB camera or RCB camera type, and its field of view is an overlapping area 7 e (area shown by hatching in the figure) overlapping with the field of vision of the radar 8, and the field of vision of the radar 8. The non-overlapping left and right non-overlapping regions 7d and 7d are arranged so as to constitute non-overlapping. In this case, the overlapping area 7e corresponds to the first predetermined area, and the area including the overlapping area 7e and the left and right non-overlapping areas 7d and 7d corresponds to the second predetermined area.

  When configured as described above, measurement data of the distance of the object by the radar 8 is calculated as the actual measurement distance Xs_o_n in the actual measurement distance calculation unit 11 described above, whereby the same effects as the vehicle control device 1 of the embodiment are obtained. You can get The measured distance Xs_o_n may be calculated by the sensor confusion method using the radar 8 and the camera 4d in the measured distance calculation unit 11 described above. Further, in the configuration of FIG. 8, LIDAR may be used instead of the radar 8.

  In the embodiment, N estimated first positions (X n — o — n, Y n — o — n) and M in the estimated position calculation unit 12 based on the image data from the left camera 4 a by the depth learning method using a depth neural network. The second estimated position (Xn_o_m, Yn_o_m) of the image data is calculated, but instead, N number 1 of first positions are obtained by the depth learning method using the depth neural network using image data from the left and right cameras 4a and 4b. The estimated position (Xn_o_n, Yn_o_n) and the M second estimated positions (Xn_o_m, Yn_o_m) may be calculated.

  Further, an estimation algorithm for calculating N first estimated positions (Xn_o_n, Yn_o_n) and M second estimated positions (Xn_o_m, Yn_o_m) in the estimated position calculation unit 12 based on the image data from the left camera 4a. Is not limited to the deep learning method using the deep neural network of the embodiment, but includes a feature amount extraction method, a pattern matching method, an optical flow method, and a method using a correlation between the height of an image and the height of an object (for example, a special feature). No. 2007-188417) may be used.

  On the other hand, although the embodiment is an example using the RGB camera type as the first camera, the first camera of the present invention is not limited to this, and any camera capable of outputting image data may be used. For example, an RCB camera may be used as the first camera, in which case yellow pixel information may be extracted by subtracting red and blue pixel information from clear pixel information.

  In addition, the embodiment is an example in which a CCC camera is used as the second camera, but the second camera of the present invention is not limited to this, and may be one having higher sensitivity than the first camera. For example, a 3CMOS camera may be used as the second camera.

  Furthermore, although the embodiment is an example in which the vehicle control device and the distance calculation device of the present invention are applied to a vehicle corresponding to left-hand traffic on the roadway, the vehicle control device and distance calculation device of the present invention are applied to the right-hand traffic on the roadway You may apply to the vehicle corresponding to. In that case, in a vehicle traveling in an environment where the traffic signal is disposed on the right, the left camera is a CCC camera capable of taking only black and white images by replacing the left and right cameras in the embodiment, and the right camera is an image of three primary colors. What is necessary is just to comprise each with the RGB camera which can image | photograph.

  On the other hand, although the embodiment is an example in which the area on the front side of the vehicle is the first predetermined area and the second predetermined area, the area on the rear side of the vehicle may be the first predetermined area and the second predetermined area. In this case, the left and right cameras may be arranged rearward.

  Further, although the embodiment is an example in which the first predetermined section is a section defined by N + M estimated distances Index_x_h, the first predetermined section is not limited thereto, and the target in the overlapping area 7c and the left overlapping area 7a What is necessary is just the distance of the object.

  Furthermore, although the embodiment is an example in which the second predetermined section is a section defined by N + M estimated lateral positions Index_y_h, the second predetermined section is not limited to this, and the second predetermined section is in the overlapping area 7c and the left overlapping area 7a. What is necessary is just the horizontal position of a target object.

  On the other hand, the embodiment is an example in which the vehicle control device 1 and the distance calculation device 1 of the present invention are applied to a four-wheeled vehicle. The present invention can also be applied to a three-wheeled vehicle and a vehicle having five or more wheels.

1 distance calculation device, vehicle control device 2 ECU (measured distance acquisition means, estimated distance calculation means, correction value calculation means, corrected distance calculation means)
3 Vehicle 4a Left camera (distance information acquisition means, image information acquisition means, first camera)
4b Right camera (distance information acquisition means, second camera)
4d camera (image information acquisition means)
7a Left non-overlapping area (part of the second predetermined area)
7c overlapping area (first predetermined area, part of second predetermined area)
7d Left and right non-overlapping area (part of the second predetermined area)
7e overlapping area (first predetermined area, part of second predetermined area)
8 Radar (distance information acquisition means)
11 Measured distance calculation unit (measured distance acquisition means)
12 Estimated position calculation unit (estimated distance calculation means)
13 Correction value calculation unit (correction value calculation means)
14 Correction position calculation unit (correction value calculation means, corrected distance calculation means)
Xs_o_n Actual distance Xn_o_n First estimated distance (estimated distance)
Xn_o_m Second estimated distance (estimated distance)
E_id_n Error signal value (deviation)
Index_x_h Estimated distance (first predetermined interval)
Index_y_h Estimated lateral position (second predetermined interval)
Kmod correction value
X_h calculated distance (distance after correction)

Claims (5)

In a distance calculation device that is mounted on a vehicle and calculates the distance between the vehicle and the object,
Distance information acquisition means for acquiring distance information;
Measured distance acquiring means for acquiring a measured distance which is a measured value of a distance to any point in the first predetermined area, using the distance information.
Image information acquisition means for acquiring image information;
An estimated distance, which is an estimated value of a distance to any point in a second predetermined area wider than the first predetermined area including all the first predetermined areas, using the image information and a predetermined estimation algorithm Estimated distance calculating means for calculating;
The measured distance of the object acquired by the measured distance acquiring means when at least a part of the object is located in the first predetermined area, and at least a part of the object is the first predetermined area The correction value for correcting the estimated distance of the object is calculated so that the absolute value of the difference between the estimated distance and the estimated distance of the object calculated by the estimated distance calculation means decreases when the position is located inside Correction value calculation means for
By correcting the estimated distance of the object calculated by the estimated distance calculating unit when the object is located in the second predetermined area including the first predetermined area, with the correction value. A corrected distance calculating means for calculating a corrected distance of the object;
A distance calculation device comprising: The distance information acquisition unit acquires the distance information by a stereo matching method using a first camera and a second camera with higher sensitivity than the first camera.
The said image information acquisition means acquires the said image information using a said 1st camera, The distance calculation apparatus of Claim 1 characterized by the above-mentioned. The distance information acquisition means acquires the distance information using one of a radar and a LIDAR,
The said image information acquisition means acquires the said image information using a camera, The distance calculation apparatus of Claim 1 characterized by the above-mentioned.   The correction value is continuous in the first predetermined section with respect to the horizontal coordinates in a two-dimensional coordinate system in which the horizontal direction with respect to the reference point of the vehicle is one coordinate and the vertical direction with respect to the reference point is one coordinate. The distance calculation device according to any one of claims 1 to 3, wherein the distance calculation device is calculated as a function value continuous in a second predetermined section with respect to the coordinate in the vertical direction. A distance calculation device according to any one of claims 1 to 4, comprising:
The first predetermined area and the second predetermined area are one of a front side and a rear side of the vehicle,
A vehicle control apparatus, comprising: controlling at least one of a driving force, a braking force, and a steering amount of the vehicle using the corrected distance of the object.