JP2016029559A - Moving body control device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To suppress increase in a processing load to enable accurate following to a target course.SOLUTION: A moving body control device comprises: a target course detecting part 26 that detects a target course from an image taken by a front camera 12; a detection point obtaining part 34 that obtains depression angles and azimuths of directions of points with respect to a camera imaging direction, regarding two points on the target course; a shape calculation part 42 that calculates, based on the depression angle and the azimuth of each of the two points, a turn curvature of the target course and a lateral deviation of a moving body to the target course; and a turning control part 48 that controls, based on the turn curvature of the target course and the lateral deviation of the moving body to the target course, a turning motion of the moving body so that it can follow the target course.SELECTED DRAWING: Figure 3

Description

  The present invention relates to a moving body control device, and more particularly to a moving body control device that controls a turning motion of a moving body that changes a moving position by a turning motion.

  In the prior art, a target course and control parameters are digitized from a front traveling scene information by a state space filter such as a Kalman filter, and the vehicle travels by position control based on the detection result (for example, Patent Document 1).

JP-A-8-261756

  In the prior art, the detection accuracy is enhanced by using estimation parameters such as a road shape model and a vehicle state quantity parameter in configuring the Kalman filter. However, the occurrence of detection errors and the processing burden increase due to their conformity state and predictive configuration. Further, in the follow-up control, the control performance deteriorates due to the detected control parameter error, and error compensation is required.

  The present invention has been made to solve the above-described problems, and an object of the present invention is to provide a mobile control apparatus that can accurately follow a target course while suppressing an increase in processing load.

  In order to achieve the above object, a moving body control device according to the present invention is a moving body control device that controls a turning motion of a moving body that changes a moving position by a turning motion, and is configured to control a moving speed of the moving body. A speed detecting means for detecting; a course detecting means for detecting a lane mark or a target course from an image taken by an imaging means for taking an image of the front or rear of the moving body; and a lane mark detected by the course detecting means or For each of a plurality of points on the target course, angle detection means for detecting a depression angle and an azimuth angle of the direction of the point with respect to a reference direction, and a depression angle and an azimuth angle detected for each of the plurality of points by the angle detection means A shape meter that calculates a turning curvature of the lane mark or the target course and a lateral deviation of the moving body with respect to the lane mark or the target course Means, a turning curvature of the lane mark or the target course calculated by the shape calculating means, a lateral deviation of the moving body with respect to the lane mark or the target course, and the moving speed detected by the speed detecting means. And a turning control means for controlling the turning motion of the moving body so as to follow the target course.

  According to the moving body control device of the present invention, the moving speed of the moving body is detected by the speed detecting means. A course detection unit detects a lane mark or a target course from an image captured by an imaging unit that captures the front or rear of the moving body.

  Then, the angle detection means detects the depression angle and the azimuth of the direction of the point with respect to the reference direction for each of the lane mark detected by the course detection means or each of a plurality of points on the target course. Based on the depression angle and azimuth angle detected for each of the plurality of points by the angle detection means by the shape calculation means, the turning curvature of the lane mark or the target course, and the moving body with respect to the lane mark or the target course Calculate the lateral deviation.

  And, by the turning control means, the turning curvature of the lane mark or the target course calculated by the shape calculating means, the lateral deviation of the moving body with respect to the lane mark or the target course, and the speed detected by the speed detecting means Based on the moving speed, the turning motion of the moving body is controlled so as to follow the target course.

  In this way, for each of a plurality of points on the lane mark or target course detected from the image, a depression angle and an azimuth angle of the direction of the point with respect to a reference direction are detected, and a depression angle detected for each of the plurality of points. And the turning curvature of the lane mark or the target course and the lateral deviation of the moving body with respect to the lane mark or the target course are calculated based on the azimuth and the turning motion of the moving body is controlled to follow the target course. By doing so, it is possible to suppress an increase in processing load and accurately follow the target course.

  The mobile control device according to the present invention calculates, for each of the plurality of points, a curvature coefficient that is an azimuth coefficient per curvature based on a depression angle of the point and a viewpoint height of the imaging unit, and the plurality of the plurality of points For each of the points, further comprising shape factor calculation means for calculating a lateral deviation coefficient that is an azimuth angle coefficient per lateral deviation based on the depression angle, azimuth angle of the point, and the viewpoint height of the imaging means, The shape calculation means includes a depression angle and an azimuth angle detected for each of the plurality of points by the angle detection means, and the curvature coefficient and the lateral deviation coefficient calculated for each of the plurality of points by the shape coefficient calculation means. The turning curvature of the lane mark or the target course and the lateral deviation of the moving body with respect to the lane mark or the target course can be calculated based on the above.

  The turning control means implements a turning yaw rate that converges a lateral deviation of the moving body with respect to the lane mark or the target course according to the turning curvature of the lane mark or the target course calculated by the shape calculating means. Thus, the turning motion of the movable body can be controlled. Further, the turning control means has a turning yaw rate that converges a lateral deviation of the moving body with respect to the lane mark or the target course according to the turning curvature of the lane mark or the target course calculated by the shape calculating means. It is possible to calculate a steering angle for realizing and to control the steering angle of the moving body.

  In addition, the mobile body control device includes a yaw rate detection unit that detects a yaw rate of the mobile body, a yaw rate of the mobile body that is detected by the yaw rate detection unit, the movement speed that is detected by the speed detection unit, and Based on the lateral deviation coefficient calculated for each of the plurality of points by the shape factor calculating means, a slip angle of the moving object is calculated, and the yaw rate of the moving object detected by the yaw rate detecting means, Based on the calculated slip angle of the moving body and the steering angle calculated by the turning control means, the yaw rate gain of the moving body with respect to the steering angle and the slip angle gain of the moving body with respect to the steering angle are calculated. And a control gain calculating means for calculating. As a result, the yaw rate gain and the slip angle gain can be adapted according to the change in characteristics of the moving body.

  In the above moving body control device, the turning curvature of the lane mark or the target course calculated by the shape calculating means, the moving speed detected by the speed detecting means, and the turning control means are calculated. Control gain calculating means for calculating a yaw rate gain of the moving body with respect to the steering angle based on the steering angle can be further included. As a result, the yaw rate gain can be adapted according to the change in characteristics of the moving body.

  The control gain calculating means calculates a slip angle gain of the moving body with respect to the steering angle according to a lateral deviation of the moving body with respect to the lane mark or target course calculated by the shape calculating means. Can do. As a result, the slip angle gain can be adapted according to the change in characteristics of the moving body.

  The above moving body control device further includes posture calculation means for calculating the posture of the moving body based on the slip angle calculated using the slip angle gain of the moving body calculated by the control gain calculation means. The turning control means calculates the steering angle based on the yaw rate gain of the moving object calculated by the control gain calculating means, controls the steering angle of the moving object, and the shape calculating means Based on the depression angle and azimuth angle detected for each of the plurality of points by the angle detection means, and the attitude of the moving body calculated by the attitude calculation means, the turning curvature of the lane mark or the target course, and the The lateral deviation of the moving body with respect to the lane mark or the target course can be calculated.

  As described above, according to the moving body control device of the present invention, the depression angle and the azimuth of the direction of the point with respect to the reference direction are detected for each of a plurality of points on the lane mark or the target course detected from the image. And calculating a turning curvature of the lane mark or the target course and a lateral deviation of the moving body with respect to the lane mark or the target course based on the depression angle and the azimuth angle detected for each of the plurality of points, By controlling the turning motion of the moving body so as to follow, it is possible to suppress an increase in processing load and to accurately follow the target course.

It is the figure shown with the visual coordinate by the angle on the basis of the camera viewpoint position and the camera imaging direction as the reference direction. It is the figure which showed the curvature yaw rate rr which follows the locus | trajectory of turning curvature ρ for following a target course, and the lateral deviation yaw rate ry which converges the lateral deviation Yer. It is the schematic which shows the structure of the vehicle control apparatus which concerns on the 1st Embodiment of this invention. It is the schematic which shows the structure of the vehicle control apparatus which concerns on the 1st Embodiment of this invention. It is a figure which shows a target course. It is a figure which shows that the movement of a vehicle is prescribed | regulated by turning motion. It is the figure which showed the front road image of the straight road. It is the figure which showed the relationship between a lateral deviation coefficient and a depression angle. It is the figure which showed the front road image of the turning road. It is the figure which showed the relationship between a curvature coefficient and a depression angle. It is a figure for demonstrating an azimuth. It is the figure which showed typically the front road shape of turning radius 400m. It is the figure which showed the example of the setting value at the time of turning circuit tracking. It is a figure for demonstrating obstruction avoidance control and follow-up control to an avoidance lane. It is a figure which shows the detection range estimation and the example of a detection of several points. It is a figure for demonstrating the method to detect the next control course shape (psi) f1, (psi) f2 in a predictive manner. It is the top view which showed the turning curvature, the lateral deviation, and the camera position. It is a flowchart which shows the content of the vehicle control processing routine in the vehicle control apparatus which concerns on the 1st Embodiment of this invention. It is a figure which shows the result of having calculated the turning curvature. It is a figure which shows the following driving | running | working result of a compound road. It is the figure which showed the driving | running | working result which added lane change tracking to the following driving | running | working of a composite road as the lane lateral position Ydrive. It is the schematic which shows the structure of the vehicle control apparatus which concerns on the 2nd Embodiment of this invention. It is a figure which shows the structure of a control block. It is a figure which shows a processing flow. It is a figure for demonstrating the method of autonomously adapting a yaw rate gain and a slip angle gain. It is a flowchart which shows the content of the vehicle control processing routine in the vehicle control apparatus which concerns on the 2nd Embodiment of this invention. It is a figure for demonstrating the method of making a yaw rate gain autonomously adapt. It is the figure which showed the vehicle body slip angle and the element regarding it. It is the figure which showed the slip angle gain adaptive coefficient code | symbol and the code | symbol of the element regarding it. It is a figure for demonstrating the method of autonomously adapting a yaw rate gain and a slip angle gain. It is a figure which shows the following driving | running | working result of a compound road at the time of using a regulation value for Ckr and Ckb. It is a figure which shows the following driving | running | working result of a compound road at the time of using CkrRT and CkbRT. It is a figure which shows the following driving | running | working result at the time of using CkrRT.

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

<Outline of First Embodiment>
<Target course shape detection (turning curvature, lateral deviation detection)>

  FIG. 1 shows visual coordinates of an angle (vertical depression angle θ, lateral azimuth angle ψ) with the camera imaging direction as a reference direction and a camera imaging direction as a reference direction. The turning curvature ρ = 1 / R and the lateral deviation Yer are calculated by the algebraic calculation based on the geometric shape from the coordinate data ψ1, ψ2 of the two detection points on the target course detected based on the lane mark or the like.

<Following target course (turning yaw rate control)>
FIG. 2 shows a curvature yaw rate rr that follows the trajectory of the turning curvature ρ for following the target course, and a lateral deviation yaw rate ry that converges the lateral deviation Yer on the turning curvature locus with a distance Ly. When the yaw rate control of the moving body is performed by the steering angle, the steering angles MaR and MaY that generate the turning curvature and the yaw rate of the lateral deviation are shown in equations (28) and (29) described later. These equations are calculated from the yaw rate steering angle gain Ckr, the vehicle speed V and the like from the characteristics of the turning motion. The position control of the moving body is performed by a turning yaw rate rm controlled by a steering angle MA obtained by adding these.

  Course shape detection and course following are realized by continuous calculation of the above algebraic expressions, and a traveling position control mechanism of the moving body is constructed.

<Configuration of vehicle control device>
As shown in FIGS. 3 and 4, the vehicle control apparatus 10 according to the first embodiment of the present invention includes a front camera 12 that images the front of the host vehicle, a vehicle speed sensor 14 that detects the speed of the host vehicle, and the like. Based on the yaw rate sensor 16 for detecting the yaw rate of the host vehicle, the front image from the front camera 12, the output from the vehicle speed sensor 14, and the output from the yaw rate sensor 16, the target course is tracked. And a computer 20 for controlling the operation of the steering mechanism 18 for controlling the steering angle.

  The front camera 12 captures an image of a target area in front of the vehicle, generates an image signal (not shown), and an A / D conversion unit (converts the image signal, which is an analog signal generated by the image capturing unit) into a digital signal. And an image memory (not shown) for temporarily storing the A / D converted image signal. The front camera 12 repeatedly captures the target area at a predetermined interval, and outputs image data indicating the front image obtained by the imaging to the computer 20.

  The vehicle speed sensor 14 repeatedly detects the vehicle speed of the host vehicle at a predetermined interval and outputs it to the computer 20.

  The yaw rate sensor 16 repeatedly detects the yaw rate of the host vehicle at predetermined intervals and outputs it to the computer 20.

  The steering mechanism 18 changes the steering angle of the host vehicle according to control by the computer 20.

  The computer 20 includes a CPU, a RAM, and a ROM that stores a program for executing a vehicle control processing routine to be described later, and is functionally configured as follows. The computer 20 includes an image acquisition unit 22, a lane mark extraction unit 24, a target course detection unit 26, a vehicle information acquisition unit 28, a detection coefficient calculation unit 30, a detection point reference value calculation unit 32, a detection point acquisition unit 34, and a shape coefficient calculation. A unit 36, a detection coefficient setting unit 38, a lane position calculation unit 40, a shape calculation unit 42, a vehicle attitude calculation unit 44, a control coefficient calculation setting unit 46, and a turning control unit 48. The lane mark extraction unit 24 and the target course detection unit 26 are an example of a course detection unit, the detection point acquisition unit 34 is an example of an angle detection unit, and the shape factor calculation unit 36 is a shape factor calculation unit. It is an example, and the shape calculation unit 42 is an example of a shape calculation unit.

  Here, the principle of the present embodiment will be described.

<Principle>
As shown in FIG. 5, in the position control of a moving body such as a vehicle, a target course determined by a prescribed road shape is set, and steering control is performed so as to follow the target course. Although the turning radius of the target course changes continuously, the behavior every moment can be shown as a steady turning as shown in FIG. As shown in FIG. 6, the movement of the vehicle is defined by a turning motion.

  The turning motion is expressed by equations (1) and (2), and the turning yaw rate r of the turning radius R is defined by the vehicle speed V. The target turning yaw rate r is controlled by a steering angle MA corresponding to a yaw rate steering gain Ckr determined as a vehicle characteristic value. However, since a lateral deviation occurs from the turning radius trajectory due to the initial posture of the vehicle, etc., it is necessary to use steering to correct it.

  The feature of this embodiment is to detect the turning curvature ρ = 1 / R and the lateral deviation Yer that are the targets of course following, and to implement the turning control that realizes the turning curvature and the turning control that converges the lateral deviation to the turning curvature locus. Do. This target detection and turning control are continuously continued. The shape of the road ahead is detected by a camera, etc., and the turning curvature and lateral deviation are calculated by algebraic calculation based on the geometrical shape of the visual coordinates based on the camera position. Vehicle position control is realized by turning control based on characteristics.

Equation of motion of steady turning 1 / R = r / V (1)
r = Ckr * MA (2)
a = V ² / R (3)

  Where a is the turning acceleration, Ckr is the yaw rate steering gain, MA is the steering angle, r is the turning yaw rate, ρ is the turning curvature (1 / R), and V is The vehicle speed.

<Target course detection>
The target course is detected from a front road image input by a camera or the like. Using the image processing, the target course is detected from the lane mark extracted by the change in the shade of the road surface. In this detection method, visual coordinate values with the camera viewpoint position as a reference value are used. The detection point is expressed as an angle with the horizontal angle as the azimuth angle ψ and the vertical direction as the depression angle θ.

<Characteristics of lateral deviation as visual coordinates>
FIG. 7 shows a road image ahead of a straight road. The lane mark extracted by the image processing and the target course detected from the lane mark are indicated by broken lines in the figure. The characteristic of lateral deviation in visual coordinates is quantified by the geometric shape as follows.

  When the turning curvature ρ = 0, the relationship between the lateral deviation azimuth angle ψy and the lateral deviation Yer of the detection point indicated by the broken line is expressed by the formula (4) by the lateral deviation coefficient Ky.

ψy = Ky * Yer (4)

  The lateral deviation coefficient Ky can be calculated from the equation (5) as a function of the camera viewpoint height He, the azimuth angle ψ, and the depression angle θ. These relationships are shown in FIG.

Ky = [ψ / tan (ψ) * tan (θ)] / He (5)

  Where Ky is a lateral deviation coefficient (deg / m), ψ is a detection point azimuth angle (deg), Yer is a lateral deviation (m, deviation from the target course), and θ is a detection It is the point depression angle (deg), and He is the camera viewpoint height (m).

<Characteristics of turning curvature as visual coordinates>
FIG. 9 shows a road image ahead of the turning road. The lane mark extracted by the image processing and the target course detected from the lane mark are indicated by broken lines in the figure. The characteristic of the turning curvature in visual coordinates is quantified by the geometric shape as follows.

  When Yer = 0, the relationship between the curvature azimuth angle ψr and the turning curvature ρ of the detection point indicated by the broken line is expressed by the equation (6) by the curvature coefficient Kr.

ψr = Kr * ρ (6)

  The curvature coefficient Kr, which is the azimuth coefficient per curvature, can be calculated from the equations (7) and (8) as a function of the depression angle θ of the detection point and the camera viewpoint height He. These relationships are shown in FIG.

Kr = tan ⁻¹ ([KrR−sqrt (KrR ² −L _θ ² )] / L _θ )
* KrR * r2d (7)
L _θ = He / tan (θ) (8)

However, Kr is a curvature coefficient (deg · m), ψr is a curvature azimuth angle (deg), KrR is a Kr calculation radius (m), and can be calculated as KrR = 400. Also, θ is the detection point depression angle (deg), He is the camera viewpoint height (m), L _θ is the depression angle distance (m), and r2d is 180 / π (deg / rad). And clearly shows the rad of the calculation formula in the trigonometric function.

  The Kr calculation radius KrR can be set based on the turning radius of the travel history detected so far. Further, as a function of the vehicle speed, for example, an average turning radius of a road that is subject to trajectory control may be a Kr calculation radius KrR. For example, when a highway is a target, an example of setting conditions is a traveling vehicle speed of 100 km. / H and a turning acceleration of 0.2 G, the turning radius 400 m determined by the equations (1), (2), and (3) may be used as the Kr calculation radius KrR.

<Azimuth angle characteristics as visual coordinates>
The relationship between the azimuth angles of the visual coordinates is shown in FIG.

ψ = ψr + ψy + ψc + ψo (9)
ψc = ψh + β (10)
ψh = sin ⁻¹ (Yv / V) * r2d (11)
Yv = ΔYer / dt (11-1)
ψo = tan ⁻¹ (Yofs / L _θ ) * r2d (12)
Yofs = Ydrive-Yline (12-1)

  However, ψc is a lane deviation angle, ψh is a vehicle body deviation angle calculated from a lateral deviation speed obtained from the Yer difference and the sampling time dt, and ψo is a lane position deviation angle that is an addition operation. , Β are vehicle body slip angles which are vehicle characteristic values. V is a vehicle speed that is a vehicle characteristic value, and Yofs is a lane position deviation that is a lateral position control target of lane coordinates. Ydrive is a lane lateral position such as lane change and avoidance tracking, and Yline is a lateral deviation of the target course.

  The azimuth angle ψ is the curvature azimuth angle ψr (formula (6)) of the target course, the lateral deviation azimuth angle ψy (formula (4)), the lane deviation angle ψc (formulas (10) and (11), as in the formula (9). )), A linear sum of the respective azimuth angles generated by the lane position deviation angle ψo (formulas (12) and (12-1)) by the lateral position control with the lateral deviation from the following course such as the lane change.

  The lane deviation angle ψc includes a vehicle body deviation angle ψh and a vehicle body slip angle β, and the vehicle body deviation angle ψh corresponds to a deviation angle from a target course determined from the speed of the lateral deviation Yer and the vehicle speed V. Is calculated from

  The vehicle speed V is detected by a vehicle speed sensor as a vehicle state quantity.

  The vehicle body slip angle β is the difference between the turning direction and the vehicle body azimuth angle caused by vehicle characteristics. The vehicle body slip angle β is calculated according to vehicle specifications or the like (formula (45) described later) and set, or the vehicle body slip angle β is approximately set to 0. Further, the vehicle body slip angle β may be estimated from the detected values of the vehicle state quantity yaw rate r and the vehicle speed V (formula (21) described later).

<Calculation method of turning curvature and lateral deviation as visual coordinates>
A method of calculating a turning curvature and a lateral deviation, which is a target of course following, which realizes position control such as a prescribed road shape and lane change, is shown.

  FIG. 12 is a diagram schematically showing the shape of a road ahead with a turning radius of 400 m. As shown in FIG. 12, two points on the target course are extracted, and detected azimuth angles ψ1, ψ2 and their depression angles θ1, θ2 are detected. As will be described later, for θ1 and θ2, a depression angle of a detection point at which detection accuracy becomes higher depending on a measurement situation or a value set in advance is used. The target course need not be on the turning course of the moving body.

  As shown in the equation (13), the turning curvature is calculated from the shape azimuth angle ψry obtained by adding the curvature azimuth angle ψr, the lateral deviation azimuth angle ψy, and the lane position deviation angle ψo having the azimuth angle component of the turning curvature and lateral deviation of the target course. Calculate the lateral deviation.

ψry = ψr + ψy + ψo (13)

  However, ψo1 and ψo2 are lane position deviation angles (deg), and ψry1 and ψry2 are shape azimuth angles (deg).

  Since the detected azimuth angles ψ1, ψ2 are values obtained by adding ψr, ψy, ψc, ψc (formulas (10) and (11)) having no azimuth component of the turning curvature and lateral deviation of the target course is ψ1, Subtract from ψ2. The lane position deviation angle ψo accompanying the lane position control is added to ψ1, ψ2 as the azimuth angle (formula (12)) of the depression angle positions θ1, θ2. Since ψc is azimuth angle information, it has the same value regardless of the depression angle θ.

  The shape azimuth angle ψry shown in the equations (14-1) and (15-1) is an azimuth angle due to the turning curvature and the lateral deviation as the tracking target.

ψry1 = ψ1-ψc + ψo1 (14-1)
ψry2 = ψ2-ψc + ψo2 (15-1)

  Here, ψ1 and ψ2 are detection point azimuth angles (deg), and ψc is a lane deviation angle (deg) (ψc = ψc1 = ψc2).

  Thus, ψc is subtracted from ψ1 and ψ2 as known values, and ψo1 and ψo2 are added to ψ1 and ψ2 as known values.

  ψry1 and ψry2 are expressed as simultaneous equations (formulas (14) and (15)) of turning curvature and lateral deviation by conversion coefficients Kr and Ky obtained as functions of θ1 and θ2, and formulas (14), When (15) is solved, equations (16) and (17) are obtained, and the turning curvature and lateral deviation can be calculated by the algebraic equations of equations (16) and (17).

ψry1 = Kr1 * ρ + Ky1 * Yer (14)
ψry2 = Kr2 * ρ + Ky2 * Yer (15)

  However, Ky and Kr are depression functions, and are calculated from the detection point depression angles θ1 and θ2 by the equations (5) and (7). Kr1 and Kr2 are curvature coefficients (deg · m), and Ky1 and Ky2 are lateral deviation coefficients (deg / m).

  FIG. 12 shows an example in which the shapes of two lane marks are detected. Line 1 has a lateral deviation of 2 m when the turning locus coincides with the target course, and the lane position deviation angle Yline of equation (12-1). Is set to 2 m to be the target course for follow-up running. In this case, (ψc = ψo = 0), ρ (= 1 / R), and Yer are calculated as follows using the algebraic expressions of equations (16) and (17).

Line1: R = 400m, Yer = 0m
Line2: R = 398m, Yer = 2m

<Calculation method of body slip angle as visual coordinates>
The vehicle body slip angle is generally calculated from vehicle specifications such as vehicle weight and vehicle speed as vehicle characteristic values, but it is also possible to estimate the vehicle body slip angle from detected values. The vehicle speed V and the turning yaw rate r are measured as vehicle estimated values, and the turning curvature of the vehicle is calculated from the equation (1).

  The two detection points on the target course that are detected to follow the trajectory are pre-reading (position, time) compared to the vehicle measurement value due to the characteristics of the detected depression angle, and are synchronized with the vehicle measurement value by filter processing etc. The calculated azimuth angles ψs1 and ψs2 are calculated.

  The azimuth angle is considered as a characteristic azimuth angle ψby (formula (18)) consisting of a vehicle body slip angle and a lateral deviation azimuth angle. The characteristic azimuth angle ψby is obtained from equations (1) and (6-1) as shown in equations (19-1) and (20-1), and ψh is calculated from equations (10) and (11). And calculated by subtracting from the synchronous azimuth angle ψs.

ψby = β + ψy (18)
ψr ′ = r / V * Kr (6-1)
ψby1 = ψs1-r / V * Kr1-ψh (19-1)
ψby2 = ψs2-r / V * Kr2-ψh (20-1)
ψby1 = β + Ky1 * Yer ′ (19)
ψby2 = β + Ky2 * Yer ′ (20)

  Where ψby1 and ψby2 are characteristic azimuth angles (deg), ψh is a vehicle body deviation angle (deg), and β is a vehicle body slip angle (deg).

  Thus, r / V * Kr is subtracted from ψs1 and ψs2 as ψr′1 and ψr′2. Further, ψh is subtracted from ψs1 and ψs2 as a known value.

  The characteristic azimuth angle ψby is a simultaneous expression of the vehicle body slip angle β and the lateral deviation Yer ′ as shown in Expressions (19) and (20). By solving Expressions (19) and (20) for β and Yer ′, (21) and (22) are obtained, and the vehicle body slip angle β can be calculated from the characteristic azimuth angles ψby1 and ψby2 and Ky1 and Ky2 by algebraic expressions using the equations (21) and (21).

  As described above, the slip angle β is calculated from the detected values such as the yaw rate and the azimuth by the algebraic expression of Expression (21).

<Follow-up control method for turning trajectory>
Based on the result of detecting the turning curvature of the target course and the lateral deviation of the vehicle from the target course from the coordinate angles of two points on the target course measured from the position of the vehicle, as shown in FIG. To control the position of the vehicle.

  The curvature yaw rate rr of the turning curvature ρ is determined as the product of the turning curvature ρ and the vehicle speed V as shown in the equation (23).

rr = ρ * V (23)

  The lateral deviation Yer caused by the turn can converge at a position on the target course in the time (formula (26)) set as the lateral deviation convergence time Ty for the turn of the lateral deviation yaw rate ry shown in the formula (24). Position control on the target course is performed by controlling the turning yaw rate rm obtained by adding these yaw rates (formula (25)).

ry = 2 * Yer / (Ty ² * V) (24)
rm = rr + ry (25)
Ty = Ly / V (26)

  However, rm is a turning yaw rate (rad / s, rm = Ckr * MA), rr is a curvature yaw rate (rad / s, rr = Ckr * MaR), and ry is a lateral deviation yaw rate (rad / s). Ry = Ckr * MaY), and Ty is a lateral deviation convergence time constant (s), which is set in advance.

  The case where the turning yaw rate rm is controlled as the product of the yaw rate steering angle gain Ckr determined by the vehicle characteristics and the steering angle MA is shown (formula (2)). From the characteristic equations (23), (24), and (2) of the turning curvature and lateral deviation turning yaw rate, the respective steering angle gains are determined as shown in equations (28) and (29). In addition, a lateral deviation speed (formula (11-1)) may be added as follow-up control compensation so that the formula (29-1) is obtained. Kdy is a speed coefficient.

MA = MaR + MaY (27)
MaR = V / Ckr * ρ (28)
MaY = 2 / (Ty ² * V * Ckr) * Yer (29)
MaY = 2 / (Ty ² * V * Ckr) * (Yer + Yv * Kdy) (29-1)

  Where Ckr is the yaw rate steering angle gain (1 / s), MA is the steering angle (rad), MaR is the curvature steering angle (rad), and MaY is the lateral deviation steering angle (rad). It is.

  The steering angle MA is calculated by algebraic expressions of equations (27), (28), and (29).

  Thus, the curvature steering angle MaR is calculated from the detected value of the turning curvature ρ, the lateral deviation steering angle MaY is calculated from the detected value of the lateral deviation Yer, the steering angle MA obtained by adding them is set, and the turning yaw rate rm is controlled. Following the target course.

  Although the example in which the curvature yaw rate rr corresponding to the turning curvature and the lateral deviation yaw rate ry corresponding to the lateral deviation are controlled by the steering angle is shown, the same position control can be performed by controlling the turning moment by the driving / braking force difference between the left and right wheels. Is possible. By replacing Ckr with the yaw rate driving control power difference gain, MaR with the curvature driving braking force difference, MaY with the lateral deviation driving braking force difference, and the steering angle MA with the driving braking force difference, the driving braking force difference is controlled to achieve the target course. Position control for following the vehicle can be realized.

<Course following>
As a coefficient setting for target course detection and course following, set lateral acceleration ar, lateral deviation convergence time Ty, and detection point correction coefficient Kt at the time of turning circuit tracking are set. The lateral acceleration ar is set according to the vehicle characteristics, running conditions, and the like. The lateral deviation convergence time Ty and the detection point correction coefficient Kt are set according to tracking conditions such as vehicle speed, lane tracking, and lane change. FIG. 13 shows an example of set values based on the set lateral acceleration ar at the time of turning circuit tracking.

The vehicle speed during turning is controlled so as not to exceed the set vehicle speed Vr calculated by the formula (30) from the set lateral acceleration ar. When the vehicle speed V exceeds Vr, deceleration is induced by speed limit or warning.
Vr = sqrt (ar * R) (30)
Lθ = V * Ty * Kt (31)

The detection point depression angle θ of the tracking locus is the depression angle L _θ calculated from the lateral deviation convergence time Ty, the correction coefficient Kt, and the vehicle speed V according to the equations (31) and (8) (the forward distance from the camera viewpoint position to the detection point). Is set from the camera height He as a reference distance. FIG. 13 shows an example calculated from the set vehicle speed Vr and the lateral deviation convergence time Ty. Although depression distance L _theta detects from the road surface in the vicinity, it is possible when V, Ty, also set correction Kt such affected by image acquisition near L _theta by inter-vehicle distance to the preceding vehicle in following travel.

  Further, the curvature coefficient may be calculated using the turning radius R calculated by the equation (30) from the traveling vehicle speed V and the set lateral acceleration ar as KrR of the equation (7).

<Lane position control>
A target course for following an arbitrary lateral position is described in the lane position deviation angle ψo (formula (12)). When taking a course by lane position control, the lane position corresponding to the turning curvature is added to the target course to determine the lane travel position at the time of turning. The course follow-up according to the sense is set by setting and selecting the turning timing and the margin based on the trajectory position at the time of turning according to the driver's sense.

For example, Ydrive is set according to the following equation (32).

Ydrive = Kdr * ρ (32)

However, Kdr is a course taking coefficient and is a set value (may be a constant, a curvature change function, or the like). As a similar effect, the lane position proportional to the steering angle MA may be set to the lane position deviation angle ψo as shown in Expression (32-1). Kdg is a steering angle proportional course taking coefficient.
ψo = Kdg * MA (32-1)

  When changing lanes by lane position control, if the driver can confirm safety by selecting lane change support, add a trajectory for changing lanes set according to the vehicle speed to the target course. , Enable driving assistance.

  For example, Ydrive is set according to the following equation (33).

Ydrive (t) = Ylc * flc * t−Slc * sin (2π * flc * t)
... (33)

  However, Ylc is a lane change width and is a set value (for example, Ylc = 4 m). Flc is a lane change frequency, which is a set value (for example, flc = 0.2 Hz). Slc is a relaxation coefficient, and is a set value (may be a constant, a time (distance) function, or the like).

  When the first avoidance assistance is performed by lane position control, an obstacle is found in the front, and when stopping is not possible and there are no obstacles in the surrounding area and safety can be ensured by avoidance, the locus for avoiding the obstacle Is added to the target course to enable safety support.

  For example, when Ydrive is a time function, Ydrive is set according to the following equation (34).

Ydrive (t) = Yev * fev * t−Sev * sin (2π * fev * t)
... (34)

  However, Yev is an avoidance width and is a set value (for example, Yev = 1 m). fev is an avoidance frequency and is a set value (fev = 0.5 Hz). Sev is a relaxation coefficient and is a set value (may be a constant, a time (distance) function, or the like).

  The above equations (32), (33), and (34) are the trajectory shape and the time function, but may be a trajectory shape based on lateral acceleration, jerk, etc. in lane position control.

  Also, when the second avoidance assistance is performed by lane position control, a front obstacle is recognized, and when braking can not be stopped and there is no obstacle around and safety can be ensured by avoidance, the obstacle, the driving lane The avoidance course is superimposed on the front image from the relative position with respect to the vehicle and the lane line (Fig. 14), the turning curvature and lateral deviation are calculated as the target course, and safety support is provided by obstacle avoidance control and follow-up control to the avoidance lane Is possible. In FIG. 14, ψ1R0 indicates the right lane mark of the host vehicle traveling lane, and ψ1L0 indicates the left lane mark. Ψ1R1 indicates the right lane mark of the right connection lane, and ψ1L1 indicates the left lane mark of the left connection lane.

<Detection of detection range and next course detection>
Although the target course is detected from the lane mark or the like, the detection rate is lowered due to a decrease in the difference in shade between the lane mark and the road surface. The following detection accuracy improvement process is performed in response to a decrease in the detection rate.

  In the target course detection method according to the present embodiment, the target course is quantified by the geometric shape based on the turning curvature and the lateral deviation, and the inverse transformation thereof is also possible. The appearance range of the lane mark is set to the ranges ψwR and ψwL estimated according to the equations (6) and (35). Within that range, the appearance probability of the lane mark is high, and the target course detected based on the detected lane mark has high reliability.

ψr = Kr * ρ (6)
ψwR = ψr + Ky ′ * (Yer + Yw)
ψwL = ψr + Ky ′ * (Yer−Yw) (35)

  Where ρ is the detected curvature (1 / m), Yer is the detected lateral deviation (m), Kr is the curvature coefficient (deg · m, equation (7)), and Ky ′ is Approximate lateral deviation coefficient (deg / m). Further, ψwR and ψwL are values (deg) that define the detection azimuth angle range, and Yw is a value (m, for example, 0.5 m) that defines the detection lateral range.

  Further, a plurality of lane marks are detected from a plurality of detection points on the road surface. The normal lane width is about 4 m, and the running lane width Ylw can be detected. When the detection probability of the following lane mark decreases, the appearance probability of a paired lane mark is high in the detection ranges ψlwR and ψlwL shown in Expression (36).

ψlwR = ψr + Ky ′ * (Yer + Ylw + Yw)
ψlwL = ψr + Ky ′ * (Yer + Ylw−Yw) (36)
Ky ′ = θ / He (5-1)

  Where Ylw is the detected lane width (m), ψlwR, ψlwL is the lane width (deg) of the detected azimuth angle range, θ is the depression angle (deg) of the detection point, and He is the camera viewpoint. Height (m).

  By detecting multiple lane marks from the detection ranges ψlwR and ψlwL, switching the lane mark to detect the target course, and integrating the lane marks of multiple lanes as shown in Fig. 14, the course shape including the lane width is reliable. To improve the detection accuracy. In addition, the target course passing through the camera position can be estimated from the geometric shape.

FIG. 15 shows an example of detection range estimation and detection of a plurality of points. These are combined to improve the detection accuracy of the target course shaped to be used directly for detection mainly turning control the L _theta shown in Equation (31) as a depression angle of the reference.

  In FIG. 16, the next control course shapes ψf1 and ψf2 are detected predictively by extending the detection point of the target course θf [1,2] in the depression direction. It is used for the preparation setting of the next control accompanying the change of the course shape and the compensation of the current control, etc. to improve the control accuracy. θf2 may be θ1, and the far detection point of the current control shape may be set as the proximity detection point of the next shape.

  Further, Equation (1) shows that the turning curvature can be estimated based on the yaw rate, vehicle speed, and the like by the vehicle state quantity detection sensor. These measurement results can be applied to the compensation of turning curvature.

<Derivation formula of curvature coefficient and lateral deviation coefficient>
FIG. 17 is a plan view showing the turning curvature, lateral deviation, and camera position. The curvature coefficient and the lateral deviation coefficient are derived based on the geometric characteristics shown in FIG. The relational expressions are shown in Expression (37) to Expression (42). The derivation formulas for the curvature coefficient are shown in Formula (43) and Formula (7-1). A formula for deriving the lateral deviation coefficient is shown in Formula (5-2).

(Relational formula for calculation of curvature coefficient and lateral deviation coefficient)
R ² = X ² + Y ² (37)
X = L _θ (38)
Y = sqrt (R ² −L _θ ² ) (39)
Yr = R−Y (40)
L _θ = He / tan (θ) (41)
_{Yy = L θ * tan (ψy} ) ··· (42)

(Derivation formula of curvature coefficient and lateral deviation coefficient)
ψr = tan ⁻¹ (Yr / Lθ) * r2d
= Tan ⁻¹ ([R−sqrt (R ² −L _θ ² )] / L _θ ) * r2d (43)
Kr = ψr / ρ
= Tan ⁻¹ ([R−sqrt (R ² −L _θ ² )] / L _θ ) * r2d * R (7-1)
Ky = ψy / Yy
= Ψy / [(He / tan (θ)) * tan (ψy)]
= [Ψy / tan (ψy) * tan (θ)] / He (5-2)

<Processing of each part of vehicle control device>
In accordance with the principle described above, in the present embodiment, the image acquisition unit 22 acquires a front image captured by the front camera 12.

  The lane mark extraction unit 24 extracts a lane mark from the front image acquired by the image acquisition unit 22. The target course detection unit 26 detects the target course based on the lane mark extracted by the lane mark extraction unit 24.

  The vehicle information acquisition unit 28 acquires the vehicle speed V detected by the vehicle speed sensor 14 and the yaw rate r detected by the yaw rate sensor 16.

  The detection coefficient calculation unit 30 calculates a Kr calculation radius KrR based on the turning radius R obtained from the turning curvature ρ calculated by the shape calculation unit 42. Based on the lateral deviation coefficients Ky1 and Ky2 calculated by the shape coefficient calculation unit 36 and the characteristic azimuth angles ψby1 and ψby2 calculated by the vehicle attitude calculation unit 44, the detection coefficient calculation unit 30 follows the above equation (21). The slip angle β is calculated. Alternatively, the vehicle body slip angle β is calculated from the vehicle body slip angle gain Ckb set by the control coefficient calculation setting unit 46 and the steering angle MA calculated by the turning control unit 48 according to the equation (45).

  The detection point reference value calculation unit 32 calculates the depression angle Lθ according to the above equation (31) based on the vehicle speed V, Ty set by the control coefficient calculation setting unit 46, and the correction coefficient Kt. In addition, the detection point reference value calculation unit 32 calculates the detection point depression angle θ according to the above equation (8) based on the depression angle distance Lθ and the camera viewpoint height He.

  The detection point acquisition unit 34 is based on the target course detected by the target course detection unit 26 and the detection point depression angle θ calculated by the detection point reference value calculation unit 32, in the vicinity of the detection point depression angle θ on the target course. Visual coordinates (θ1, ψ1) and (θ2, ψ2) are acquired for two points.

  The shape factor calculation unit 36 calculates the lateral deviation according to the above formula (5) based on the camera viewpoint height He for each of the two points (θ1, ψ1) and (θ2, ψ2) acquired by the detection point acquisition unit 34. The coefficients Ky1 and Ky2 are calculated. Based on the Kr calculation radius KrR calculated by the detection coefficient calculation unit 30 for each of the two points (θ1, ψ1) and (θ2, ψ2) acquired by the detection point acquisition unit 34, the shape factor calculation unit 36 The curvature coefficients Kr1 and Kr2 are calculated according to the equations (7) and (8).

  The detection coefficient setting unit 38 sets the lateral acceleration ar, the lateral deviation convergence time Ty, and the detection point correction coefficient Kt, and sets the initial value of the Kr calculation radius KrR.

  For example, when the course is taken as the lane position control, the lane position computing unit 40 is based on the turning curvature ρ calculated by the shape computing unit 42 and the course taking coefficient Kdr according to the above equation (32). Ydrive is calculated. In addition, the lane position calculation unit 40 calculates the above formulas (12) and (12) for each of the two points (θ1, ψ1) and (θ2, ψ2) acquired by the detection point acquisition unit 34 based on the lane lateral position Ydrive. In accordance with 12-1), the lane position deviation angles ψo1 and ψo2 are calculated.

  The shape calculation unit 42 includes a lane deviation angle ψc calculated by the vehicle attitude calculation unit 44 and a lane position calculation unit for each of the two points (θ1, ψ1) and (θ2, ψ2) acquired by the detection point acquisition unit 34. Based on the lane position deviation angles ψo1 and ψo2 calculated by 40, the shape azimuth angles ψry1 and ψry2 are calculated according to the above formulas (14-1) and (15-1). Based on the shape azimuth angles ψry1 and ψry2 and the curvature coefficients Kr1 and Kr2 and the lateral deviation coefficients Ky1 and Ky2 calculated by the shape coefficient calculation unit 36, the shape calculation unit 42 calculates the above formulas (16) and (17). Then, the turning curvature ρ and the lateral deviation Yer are calculated. The shape calculation unit 42 calculates road shape data such as the lane width Ylw based on the lane mark extracted by the lane mark extraction unit 24.

  The vehicle attitude calculation unit 44 calculates the vehicle body deviation angle ψh according to the above equation (11) based on the difference ΔYer of the lateral deviation Yer calculated by the shape calculation unit 42 from the previous value and the vehicle speed V. The vehicle attitude calculation unit 44 calculates the lane deviation angle ψc according to the above equation (10) based on the calculated vehicle body deviation angle ψh and the vehicle body slip angle β. The vehicle attitude calculation unit 44 calculates the synchronous azimuth angles ψs1 and ψs2 in which the azimuth angles ψ1 and ψ2 are synchronized with the vehicle measurement values by filter processing and the like, and the curvature coefficients Kr1 and Kr2 calculated by the shape factor calculation unit 36. Based on the vehicle body deviation angle ψh, the vehicle speed V, and the yaw rate r, the characteristic azimuth angles ψby1 and ψby2 are calculated according to the equations (19-1) and (20-1).

  The control coefficient calculation setting unit 46 calculates the yaw rate steering angle gain Ckr and the vehicle body slip angle steering angle gain Ckb according to the following equations (44) and (45-1).

Ckr = 1 / (1 + A * V ² ) * V / L * 1 / Ng (44)
β = Ckb * MA (45)
Ckb = (1 + A ′ * V ² ) / (1 + A * V ² ) * lr / L * 1 / Ng * r2d
... (45-1)
A = − (m * (lf * Cf−lr * Cr)) / (2 * L ² * Cf * Cr) (46)
A ′ = − (m * lf) / (2 * L * lr * Cr) (47)

  Here, Cf is a front wheel cornering power, Cr is a rear wheel cornering power, and L is a distance between the front wheel shaft and the rear wheel shaft. Also, If is the distance between the front wheel axis and the center of gravity, lr is the distance between the rear wheel axis and the center of gravity, and m is the mass of the vehicle. Ng is a steering gear ratio. The above various values are calculated based on vehicle specifications, stored in memory as map data of vehicle speed, and corrected.

  The control coefficient calculation setting unit 46 sets the lateral deviation convergence time constant Ty and the correction coefficient Kt according to the following condition of the lane position calculation unit 40.

  The turning control unit 48 includes a turning curvature ρ and a lateral deviation Yer calculated by the shape calculating unit 42, a yaw rate steering angle gain Ckr calculated by the control coefficient calculation setting unit 46, and a set lateral deviation convergence time constant Ty, Based on the vehicle speed V, the steering angle MA is calculated according to the above equations (27) to (29), and the steering mechanism 18 is controlled according to the calculated steering angle MA.

<Operation of vehicle control device>
Next, the operation of the vehicle control device 10 according to the first embodiment will be described. While the vehicle equipped with the vehicle control device 10 is traveling, the computer 20 executes a vehicle control processing routine shown in FIG.

  First, in step 100, the detection coefficient setting unit 38 sets the lateral acceleration ar, the lateral deviation convergence time Ty, and the detection point correction coefficient Kt, and sets the initial value of the Kr calculation radius KrR.

  In step 102, the lane position calculation unit 40, for example, when taking a course as lane position control, is based on the turning curvature ρ previously calculated in step 118, which will be described later, and a preset course acquisition coefficient Kdr. The lane lateral position Ydrive is calculated according to the above equation (32). In addition, the lane position calculation unit 40 calculates the above formulas (12) and (12) for each of the two points (θ1, ψ1) and (θ2, ψ2) acquired by the detection point acquisition unit 34 based on the lane lateral position Ydrive. In accordance with 12-1), the lane position deviation angles ψo1 and ψo2 are calculated.

  In step 104, the vehicle information acquisition unit 28 acquires the vehicle speed V detected by the vehicle speed sensor 14 and the yaw rate r detected by the yaw rate sensor 16.

  In step 106, the detection coefficient calculation unit 30 calculates a Kr calculation radius KrR based on the turning radius R obtained from the turning curvature ρ previously calculated in step 118. Based on the lateral deviation coefficients Ky1 and Ky2 previously calculated in step 116, which will be described later, and the characteristic azimuth angles ψby1, ψby2 previously calculated in step 120, which will be described later, the detection coefficient calculation unit 30 follows the above equation (21). The slip angle β is calculated. Further, Ty and Kt are compensated for according to the lane lateral position Ydrive set in step 102 and the vehicle speed V acquired in step 104.

  In step 108, the image acquisition unit 22 acquires a front image captured by the front camera 12.

  In step 110, the lane mark extraction unit 24 extracts a lane mark from the front image acquired in step 108. And the target course detection part 26 detects a target course based on the extracted lane mark.

  In step 112, the detection point reference value calculation unit 32 calculates the depression angle Lθ according to the above equation (31) based on the vehicle speed V acquired in step 104 and Ty and Kt set in step 100. To do. In addition, the detection point reference value calculation unit 32 calculates the detection point depression angle θ according to the above equation (8) based on the depression angle distance Lθ and the camera viewpoint height He.

  In step 114, the detection point acquisition unit 34 determines two points near the detection point depression angle θ on the target course based on the target course detected in step 110 and the detection point depression angle θ calculated in step 112. The visual coordinates (θ1, ψ1), (θ2, ψ2) are acquired.

  In step 116, the shape factor calculation unit 36 calculates the two points (θ1, ψ1) and (θ2, ψ2) acquired in step 114 according to the above formula (5) based on the camera viewpoint height He. Lateral deviation coefficients Ky1 and Ky2 are calculated. Based on the Kr calculation radius KrR calculated in step 106 for each of the two points (θ1, ψ1) and (θ2, ψ2) acquired in step 114, the shape factor calculation unit 36 calculates the equation (7). The curvature coefficients Kr1 and Kr2 are calculated according to the equation (8).

  In step 118, the shape calculation unit 42 determines the lane deviation angle ψc previously calculated in step 120 for each of the two points (θ1, ψ1) and (θ2, ψ2) acquired in step 114, and the step 102. Based on the calculated lane position deviation angles ψo1 and ψo2, the shape azimuth angles ψry1 and ψry2 are calculated according to the above formulas (14-1) and (15-1). Based on the shape azimuth angles ψ ry 1 and ψ ry 2 and the lateral deviation coefficients Ky 1 and Ky 2 computed in step 106, the shape computation unit 42 turns the turning curvature ρ and the lateral deviation according to the above formulas (16) and (17). Yer is calculated. The shape calculation unit 42 calculates the lane width Ylw based on the lane mark extracted in step 110.

  In step 120, the vehicle attitude calculation unit 44 follows the equation (11) based on the difference ΔYer between the lateral deviation Yer calculated in step 118 and the previous value and the vehicle speed V acquired in step 104. Then, the vehicle body deviation angle ψh is calculated. The vehicle posture calculation unit 44 calculates the lane deviation angle ψc according to the above equation (10) based on the vehicle body slip angle β. The vehicle attitude calculation unit 44 synchronizes azimuth angles ψ 1 and ψ s 2 that synchronize the azimuth angles ψ 1 and 2 with the vehicle measurement values by filtering, the curvature coefficients Kr 1 and Kr 2 calculated in step 116, and the calculated vehicle body Based on the deviation angle ψh, the vehicle speed V, and the yaw rate r, the characteristic azimuth angles ψby1 and ψby2 are calculated according to the above equations (19-1) and (20-1).

  In step 122, the control coefficient calculation setting unit 46 calculates the yaw rate steering angle gain Ckr according to the above equation (44).

  In step 124, the turning control unit 48 determines the turning curvature ρ and the lateral deviation Yer calculated in step 118, the yaw rate steering angle gain Ckr and the lateral deviation convergence time constant Ty calculated in step 122, the vehicle speed V, and the like. Based on the above, the steering angle MA is calculated according to the above equations (27) to (29), the steering mechanism 18 is controlled according to the calculated steering angle MA, and the process returns to step 102.

<Example>
Next, examples of actual detection results and trajectory tracking control will be shown.

<Detection result of turning curvature>
FIG. 19 shows a travel of 100 m from a straight road 50 m in the X-axis direction and moves on a course of steady turning 50 m with a turning radius of 400 m, using two azimuth angles of 4 and 6 degrees from the coordinate position. The result of calculating the turning curvature is shown. However, as detection conditions, KrR = 400 m, θ1 = 4 degrees, θ2 = 6 degrees, V = 100 km / h, and detection period = 60 Hz were used.

  As shown in FIG. 19, the detected curvature in the turning circuit and steady turning coincides with the course curvature of the target course. The relaxation part detects the turning curvature in advance by the depression angle θ, and can continuously detect a continuous curvature change. Moreover, the curvature change of a clothoid connection part is relieve | moderated and the lateral acceleration change at the time of course following is reduced.

<Control result of following steering>
FIG. 20 and FIG. 21 show the results of running following a compound road of a straight road and a turning circuit as a target course. The shape of the target course consists of a straight road S, relaxation (clothoid) C, and steady turning (R400) R as shown in the figure. The vehicle detects a turning curvature 1 / R and a lateral deviation Yer from two azimuth angles ψ1, ψ2, and depression angles θ1, θ2 on a target course measured by a camera from the road ahead, and a steering angle MaR for generating each turning yaw rate, MaY and its added steering angle MA are calculated, and the follow-up traveling by the yaw rate rm generated by the yaw rate gain Ckr is continuously repeated. As control conditions, θ1 = 4 degrees, θ2 = 6 degrees, Ty = 0.4 sec, V = 100 km / h, detection azimuth angle accuracy = 0.05 degrees, and control cycle = 10 Hz were used.

  FIG. 20 shows the result of following the complex road. The detected lateral deviation Yer varies in the measurement width according to the detected azimuth angle accuracy, but the camera lane deviation from the target course of the camera position is controlled to be 10 mm or less. The turning curvature of the target course (following course curvature) matches the detected curvature, and is detected correctly.

  FIG. 21 shows a traveling result obtained by adding the setting value exemplified in the equation (33) to the following traveling of the composite road, with the lane change tracking as the lane lateral position Ydrive. The camera lane deviation (displayed as the lateral deviation Yer from the tracking shape of the target course) follows the lane change target course with high accuracy. The steering angle MA is controlled by superimposing lane change tracking on road track tracking (see FIG. 20).

  As described above, according to the vehicle control device according to the first embodiment, for each of the two points on the target course detected from the image, the depression angle of the direction of the point with respect to the camera imaging direction from the camera viewpoint And the azimuth angle are detected, the turning curvature of the target course and the lateral deviation of the vehicle with respect to the target course are calculated based on the depression angle and the azimuth angle detected for each of the two points, and the vehicle follows the target course. By controlling the turning motion of the vehicle, an increase in processing load can be suppressed and the target course can be accurately followed.

  In addition, position control of the host vehicle is performed by following the target course. The turning curvature and the lateral deviation are detected as the target course shape, and the turning control for realizing the turning curvature and the turning control for converging the lateral deviation to the turning curvature locus are performed. The course is followed by continuing the shape detection and turning control. Thus, the position control of the host vehicle can be realized simply and with high accuracy by detecting the target course shape as the turning curvature and lateral deviation from moment to moment and controlling the behavior defined as the turning motion.

  Next, a second embodiment will be described. In addition, the same code | symbol is attached | subjected about the part which becomes the same structure as 1st Embodiment, and description is abbreviate | omitted.

<Outline of Second Embodiment>
In the first embodiment, vehicle control is performed based on control characteristics and control coefficients set in advance. If the characteristics of the vehicle being controlled, such as weight, changes, the response characteristics of the controlled object will differ from the assumed conditions, resulting in an excess or deficiency of the controlled variable, resulting in a decrease in control accuracy and an inability to perform optimal position control. There is a case.

  The target course shape is detected from the relative posture of the course and the vehicle. If the vehicle posture is not correctly detected, a measurement error of the target course shape occurs, and the control accuracy decreases. In the first embodiment, the posture of the vehicle, for example, the vehicle body slip angle is estimated from the slip angle gain with respect to the steering angle. The accuracy of the posture detection decreases due to the accuracy of the slip angle gain value and the change of the slip angle gain accompanying the change in the characteristics of the controlled object, and the accuracy of the control decreases.

  Therefore, in the present embodiment, in the vehicle control device that controls the position of the vehicle for following the target course, the response characteristic of the vehicle is detected or detected by detecting or estimating the behavior of the vehicle with respect to the control amount in real time. Estimate and autonomously adapt by sequentially updating and resetting control characteristics to optimize control accuracy and stability.

<Configuration of vehicle control device>
As shown in FIG. 22, the computer 220 of the vehicle control apparatus 210 according to the second embodiment of the present invention includes an image acquisition unit 22, a lane mark extraction unit 24, a target course detection unit 26, a vehicle information acquisition unit 28, Detection coefficient calculation unit 30, detection point reference value calculation unit 32, detection point acquisition unit 34, shape coefficient calculation unit 36, detection coefficient setting unit 38, lane position calculation unit 40, shape calculation unit 42, vehicle attitude calculation unit 44, control A coefficient setting unit 246, a control coefficient calculation unit 250, and a turning control unit 48 are provided. The detection coefficient calculation unit 30 is an example of an attitude calculation unit, and the control coefficient calculation unit 250 is an example of a control gain calculation unit.

  Here, the principle of the present embodiment will be described.

<Principle>
The target course shape to be followed is photographed from the front road image by the front camera 12, and the road shape is calculated as the steady turning curvature ρ and the lateral deviation Yer from the image data. As vehicle motion information, the vehicle speed V and the yaw rate r are measured by the vehicle speed sensor 14 and the yaw rate sensor 16. The vehicle is controlled by the turning yaw rate control, and the turning yaw rate control is performed by the steering angle MA obtained by adding the MaR set from the turning curvature ρ and the MaY set from the lateral deviation Yer (see FIGS. 23 and 24).

  The block diagram shown in FIG. 23 and the processing flow diagram shown in FIG. 24 are common to the first embodiment, and the corresponding expressions of each block in FIG. 23 are shown below.

  ρout is calculated by equation (16), etc., Yout is calculated by equation (17), etc., KdR is calculated by equation (28), and KdY is calculated by equation (29). In addition, Ty · V is calculated by equations (26) and (31), Ckr is calculated by equation (44) and the like, and Ckb is calculated by equation (45-1) and the like. s represents a differential operator, and 1 / s represents an integral operator.

  The yaw rate gain Ckr with respect to the steering angle and the slip angle gain Ckb with respect to the steering angle change due to changes in characteristics such as the vehicle weight of the vehicle to be controlled. , Stability decreases.

  In the present embodiment, optimum control characteristics are reset by detecting the yaw rate gain Ckr and slip angle gain Ckb in real time.

  It is obvious that the yaw rate gain and the slip angle gain may be measured in real time, but it is difficult to use the real time measurement value due to a time delay due to processing and mutual interference of detection elements. In the present embodiment, these problems are solved by optimizing the processing, and real-time measurement is possible.

  Hereinafter, in order to distinguish from Ckr and Ckb described in the first embodiment, the detection values in real time are denoted as CkrRT and CkbRT.

<Processing of each part of vehicle control device>
In accordance with the principle described above, in the present embodiment, the control coefficient calculation unit 250 has the lateral deviation coefficients Ky1 and Ky2 calculated by the shape coefficient calculation unit 36 and the characteristic azimuth angle ψby1 calculated by the vehicle attitude calculation unit 44. , Ψby 2, the slip angle β is calculated according to the above equation (21) (denoted as βRT).

  The control coefficient calculation unit 250 uses the turning yaw rate r measured by the yaw rate sensor 16, the calculated slip angle βRT, and the steering angle MA generated by the turning control unit 48, and the following equations (101), (102) Thus, autonomous adaptation is performed by calculating the yaw rate gain CkrRT and the slip angle gain CkbRT (see FIG. 25).

CkrRT = r / MA (101)
CkbRT = βRT / MA (102)

  In the first embodiment, it is described in a form that is directly used as the slip angle. Therefore, although the accuracy of βRT is restricted by the accuracy and delay of the synchronous azimuth values ψs1, ψs2, the slip angle gain CkbRT By setting as the slip angle of the steering angle function, it becomes easy to estimate the slip angle with high accuracy and to set the delay characteristic. In addition, the required accuracy of ψs1 and ψs2 for βRT calculation can be lowered, and stable detection under various driving conditions becomes possible. The detection of the slip angle is important for highly accurate position detection of the vehicle. In the shape detection using the azimuth angle as in the present embodiment, the slip angle can be directly detected from the image data, and stable autonomous adaptation is realized.

  The control coefficient setting unit 246 uses the yaw rate gain CkrRT and slip angle gain CkbRT calculated by the control coefficient calculation unit 250 to reset the yaw rate gain Ckr and slip angle gain Ckb. Similarly to the control coefficient calculation setting unit 46, the control coefficient setting unit 246 sets the lateral deviation convergence time constant Ty and the correction coefficient Kt according to the following condition of the lane position calculation unit 40.

  The turning control unit 48 includes a turning curvature ρ and a lateral deviation Yer calculated by the shape calculating unit 42, a yaw rate steering angle gain CkrRT calculated by the control coefficient setting unit 246, a set lateral deviation convergence time constant Ty, and a vehicle speed. Based on V, the steering angle MA is calculated according to the above formula (27), (28), (29) or (29-1), and the steering mechanism 18 is controlled according to the calculated steering angle MA. Thus, by using the reset CkrRT, the steering angles MA, MaR, and MaY for position control are always optimized.

  The detection coefficient calculation unit 30 calculates the vehicle body slip angle β from the vehicle body slip angle steering angle gain CkbRT set by the control coefficient setting unit 246 and the steering angle MA calculated by the turning control unit 48 according to the above equation (45). Calculate.

  The vehicle attitude calculation unit 44 calculates the vehicle body deviation angle ψh according to the above equation (11) based on the difference ΔYer of the lateral deviation Yer calculated by the shape calculation unit 42 from the previous value and the vehicle speed V. Based on the calculated vehicle body deviation angle ψh and the vehicle body slip angle β calculated by the detection coefficient calculation unit 30, the vehicle posture calculation unit 44 calculates the lane deviation angle ψc according to the above equation (10).

  The shape calculation unit 42 includes a lane deviation angle ψc calculated by the vehicle attitude calculation unit 44 and a lane position calculation unit for each of the two points (θ1, ψ1) and (θ2, ψ2) acquired by the detection point acquisition unit 34. Based on the lane position deviation angles ψo1 and ψo2 calculated by 40, the shape azimuth angles ψry1 and ψry2 are calculated according to the above formulas (14-1) and (15-1).

  Thus, by using the reset CkbRT, the azimuth angles ψc, ψry1, and ψry1 for detecting the target course shape are always optimized.

<Operation of vehicle control device>
Next, the operation of the vehicle control device 210 according to the second embodiment will be described. A vehicle control processing routine shown in FIG. 26 is executed in the computer 220 while the vehicle equipped with the vehicle control device 210 is traveling. In addition, about the process similar to 1st Embodiment, the same code | symbol is attached | subjected and detailed description is abbreviate | omitted.

  First, in step 100, the detection coefficient setting unit 38 sets the lateral acceleration ar, the lateral deviation convergence time Ty, and the detection point correction coefficient Kt, and sets the initial value of the Kr calculation radius KrR. In step 102, the lane position calculation unit 40 calculates the lane lateral position Ydrive according to the above equation (32) based on the turning curvature ρ previously calculated in step 118 and the preset course taking coefficient Kdr. To do. In addition, the lane position calculation unit 40 calculates the above formulas (12) and (12) for each of the two points (θ1, ψ1) and (θ2, ψ2) acquired by the detection point acquisition unit 34 based on the lane lateral position Ydrive. In accordance with 12-1), the lane position deviation angles ψo1 and ψo2 are calculated.

  In step 104, the vehicle information acquisition unit 28 acquires the vehicle speed V detected by the vehicle speed sensor 14 and the yaw rate r detected by the yaw rate sensor 16.

  In step 106, the detection coefficient calculation unit 30 calculates a Kr calculation radius KrR based on the turning radius R obtained from the turning curvature ρ previously calculated in step 118. The detection coefficient calculation unit 30 calculates a vehicle body slip angle β from the vehicle body slip angle gain CkbRT previously set in step 272 described later and the steering angle MA previously calculated in step 124 according to the above equation (45). Further, Ty and Kt are compensated for according to the lane lateral position Ydrive set in step 102 and the vehicle speed V acquired in step 104.

  In step 108, the image acquisition unit 22 acquires a front image captured by the front camera 12.

  In step 110, the lane mark extraction unit 24 extracts a lane mark from the front image acquired in step 108. And the target course detection part 26 detects a target course based on the extracted lane mark.

  In step 112, the detection point reference value calculation unit 32 calculates the depression angle Lθ according to the above equation (31) based on the vehicle speed V acquired in step 104 and Ty and Kt set in step 100. To do. In addition, the detection point reference value calculation unit 32 calculates the detection point depression angle θ according to the above equation (8) based on the depression angle distance Lθ and the camera viewpoint height He.

  In step 114, the detection point acquisition unit 34 determines two points near the detection point depression angle θ on the target course based on the target course detected in step 110 and the detection point depression angle θ calculated in step 112. The visual coordinates (θ1, ψ1), (θ2, ψ2) are acquired.

  In step 116, the shape factor calculation unit 36 calculates the two points (θ1, ψ1) and (θ2, ψ2) acquired in step 114 according to the above formula (5) based on the camera viewpoint height He. Lateral deviation coefficients Ky1 and Ky2 are calculated. Based on the Kr calculation radius KrR calculated in step 106 for each of the two points (θ1, ψ1) and (θ2, ψ2) acquired in step 114, the shape factor calculation unit 36 calculates the equation (7). The curvature coefficients Kr1 and Kr2 are calculated according to the equation (8).

  In step 118, the shape calculation unit 42 determines the lane deviation angle ψc previously calculated in step 120 for each of the two points (θ1, ψ1) and (θ2, ψ2) acquired in step 114, and the step 102. Based on the calculated lane position deviation angles ψo1 and ψo2, the shape azimuth angles ψry1 and ψry2 are calculated according to the above formulas (14-1) and (15-1). Based on the shape azimuth angles ψ ry 1 and ψ ry 2 and the lateral deviation coefficients Ky 1 and Ky 2 computed in step 106, the shape computation unit 42 turns the turning curvature ρ and the lateral deviation according to the above formulas (16) and (17). Yer is calculated. The shape calculation unit 42 calculates the lane width Ylw based on the lane mark extracted in step 110.

  In step 120, the vehicle attitude calculation unit 44 follows the equation (11) based on the difference ΔYer between the lateral deviation Yer calculated in step 118 and the previous value and the vehicle speed V acquired in step 104. Then, the vehicle body deviation angle ψh is calculated. The vehicle attitude calculation unit 44 calculates the lane deviation angle ψc according to the above equation (10) based on the vehicle body slip angle β calculated in step 106. The vehicle attitude calculation unit 44 synchronizes azimuth angles ψ 1 and ψ s 2 that synchronize the azimuth angles ψ 1 and 2 with the vehicle measurement values by filtering, the curvature coefficients Kr 1 and Kr 2 calculated in step 116, and the calculated vehicle body Based on the deviation angle ψh, the vehicle speed V, and the yaw rate r, the characteristic azimuth angles ψby1 and ψby2 are calculated according to the above equations (19-1) and (20-1).

  In step 124, the turning control unit 48 calculates the turning curvature ρ and the lateral deviation Yer calculated in step 118, the yaw rate steering angle gain CkrRT previously calculated in step 272, and the set lateral deviation convergence time constant Ty. Based on the vehicle speed V, the steering angle MA is calculated according to the above formula (27), (28), (29) or (29-1), and the steering mechanism 18 is controlled according to the calculated steering angle MA.

  In step 270, the control coefficient calculation unit 250 calculates the above formula (21) based on the lateral deviation coefficients Ky1 and Ky2 calculated in step 116 and the characteristic azimuth angles ψby1 and ψby2 calculated in step 120. ) To calculate the slip angle βRT. The control coefficient calculation unit 250 uses the turning yaw rate r acquired in step 104, the calculated slip angle βRT, and the steering angle MA generated by the turning control unit 48, and uses the following equations (101) and (102). Therefore, the yaw rate gain CkrRT and the slip angle gain CkbRT are calculated.

  In step 272, the yaw rate gain Ckr and slip angle gain Ckb are reset using the yaw rate gain CkrRT and slip angle gain CkbRT calculated in step 270, and the process returns to step 102.

  As described above, according to the vehicle control apparatus according to the second embodiment, in vehicle control, steering that becomes a control target characteristic value from a measured value of a turning yaw rate, image data obtained by imaging a front road environment, or the like. By detecting the yaw rate gain with respect to the angle and the vehicle body slip angle gain with respect to the steering angle, which is the environment detection characteristic value, in real time, the control characteristics can be adjusted appropriately even if the specifications such as the weight of the controlled vehicle change The accuracy and stability of the control can be maintained while maintaining the state. In addition, by resetting the optimum control characteristics in real time, the vehicle control characteristics corresponding to the characteristic changes of the vehicle to be controlled are realized. In addition, optimal control characteristics can be realized under conditions other than those assumed in advance.

<Third Embodiment>
Next, a third embodiment will be described. In addition, since the structure of the vehicle control apparatus which concerns on 3rd Embodiment becomes a structure similar to 2nd Embodiment, it attaches | subjects the same code | symbol and abbreviate | omits description.

  Here, the principle of the present embodiment will be described.

<Principle>
In the present embodiment, a slip angle gain set as shown in the above equation (45-1) can be applied, and it adapts to a change in yaw rate gain due to a change in vehicle characteristics.

  In the configuration in which the vehicle speed V is measured by the vehicle speed sensor 14, the road image is captured by the front camera 12 and the road shape is captured as the image data, as described in the first embodiment, the above equation (16 ) And (17), the yaw rate gain CkrRT is adaptively detected from the turning curvature ρ of the target course shape, the lateral deviation Yer, the vehicle speed V, and the steering angles MaR and MaY set from them.

  The steering angle is set as the above equation (28) according to the turning curvature, and from this equation, the yaw rate gain becomes the equation (103).

MaR = V / Ckr * ρ (28)
Ckr = V * ρ / MaR (103)

  When Ckr differs from the set value due to a change in vehicle characteristics, a lateral deviation occurs as a control error, and MaY is set according to equation (29). CkrRT is set by Equation (104) so that MaY corresponding to the control error converges. As CkrRT converges to a positive value corresponding to changes in vehicle characteristics, MaY * KyRT decreases and can be autonomously adapted. The yaw rate gain adaptation coefficient KyRt may be set to an appropriate value according to the adaptation condition.

MaY = 2 / (Ty ² * V * Ckr) * Yer (29)
CkrRT = V * ρ / (MaR + MaY * KyRT) (104)

<Processing of each part of vehicle control device>
The control coefficient calculation unit 250 uses the vehicle speed V measured by the vehicle speed sensor 14, the turning curvature ρ calculated by the shape calculation unit 42, and the steering angle MA generated by the turning control unit 48. Therefore, autonomous adaptation is performed by calculating the yaw rate gain CkrRT (see FIG. 27).

  The control coefficient setting unit 246 uses the yaw rate gain CkrRT calculated by the control coefficient calculation unit 250 to reset the yaw rate gain Ckr. Similarly to the control coefficient calculation setting unit 46, the control coefficient setting unit 246 sets the lateral deviation convergence time constant Ty and the correction coefficient Kt according to the following condition of the lane position calculation unit 40. Control coefficient setting unit 246 sets a predetermined value as slip angle gain Ckb.

  In addition, about the other structure and effect | action of the vehicle control apparatus which concern on 3rd Embodiment, since it is the same as that of 2nd Embodiment, description is abbreviate | omitted.

  As described above, according to the vehicle control device according to the third embodiment, the yaw rate gain can be adapted in accordance with the change in the characteristics of the vehicle.

<Fourth embodiment>
Next, a fourth embodiment will be described. In addition, since the structure of the vehicle control apparatus which concerns on 4th Embodiment becomes a structure similar to 2nd Embodiment, it attaches | subjects the same code | symbol and abbreviate | omits description.

  Here, the principle of the present embodiment will be described.

<Principle>
The present embodiment adapts to changes in the yaw rate gain and slip angle gain due to changes in vehicle characteristics.

  In the configuration in which the vehicle speed V is measured by the vehicle speed sensor 14, the road image is captured by the front camera 12 and the road shape is captured as the image data, as described in the first embodiment, the above equation (16 The vehicle body slip angle gain CkbRT and the yaw rate gain CkrRT are applied from the turning curvature ρ of the target course shape detected in accordance with (17), the lateral deviation Yer, the vehicle speed V, and the steering angles MaR and MaY set from them.

  The yaw rate gain is autonomously adapted by the above equation (104).

  The vehicle body slip angle corresponds to the difference between the vehicle body azimuth angle and the turning direction during turning.

  The turning curvature and lateral deviation of the target course shape can be detected correctly when the turning direction and the vehicle body azimuth coincide with each other, and the detected angle is corrected by the vehicle body slip angle gain Ckb or the like in order to prevent errors due to the vehicle body slip angle (the above formula (10 ) (13) (14-1) (15-2) (14) (15) (16) (17) (45) etc.).

  If the actual vehicle body slip angle gain has a set value and an error due to a change in vehicle characteristics, a turning curvature and lateral deviation detection error occurs. Therefore, in the present embodiment, the vehicle body slip angle gain is adapted by converging the lateral deviation.

  The azimuth angle difference between the vehicle body direction at the time of turning and the turning direction of the target course is defined as a vehicle body slip angle. The turning outer direction is represented by an outward slip angle and a plus sign (+), and the reverse turning inner direction is taken as an inward slip angle and a minus sign (−). The slip angle gain also uses the same sign as the slip angle and depends on the turning direction and the steering angle MA.

  Also, consider a coordinate system in which the turning direction and steering angle MA associated with a right turn are plus (+), and the lateral deviation Yer detected by the slip angle is also the right direction (+). If the slip angle is not correctly estimated, Yer is detected as shown in FIG. 28 due to the slip angle error.

  From the above relationship, the condition for the vehicle body slip angle to be inward is the value obtained by adding the reverse sign (minus) of the steering angle MA to the lateral deviation Yer, and the slip angle deviation sYer (formula (105)) is plus + Become. Conversely, the vehicle body slip angle is outward when the equivalence is negative (-).

  The lateral deviation due to the vehicle body slip angle is normally corrected, but the lateral deviation is erroneously detected due to the error of the slip angle gain Ckb. CkbRT is autonomously adapted by equations (106) and (107) so that the slip angle deviation sYer in equation (105) is equal to or less than the specified slip angle deviation Yebeta (see FIG. 29). The slip angle deviation prescribed value Yebeta and the slip angle gain adaptation coefficient Ckb_Rup are set to appropriate values according to the adaptation conditions.

sYer = Yer * sign (−MA) (105)

  When the judgment formula (106) in which the vehicle body slip angle is outward is established, CkbRT is increased to the (+) side according to the formula (107).

sYer> Yebeta (106)
CkbRT = CkbRT + Ckb_Rup (107)

  On the other hand, when the judgment formula (108) in which the vehicle body slip angle is inward is established, CkbRT is increased to the (−) side according to the formula (109).

sYer <(−Yebeta) (108)
CkbRT = CkbRT − Ckb_Rup (109)

  As CkbRT set in the above equations (107) and (109) converges to a positive value corresponding to a change in vehicle characteristics, sYer decreases and can be autonomously adapted.

  CkbRT is reset as appropriate. For example, the resetting may be performed when the vehicle goes straight (when the lateral acceleration becomes a specified value or less).

<Processing of each part of vehicle control device>
The control coefficient calculation unit 250 uses the vehicle speed V measured by the vehicle speed sensor 14, the turning curvature ρ calculated by the shape calculation unit 42, and the steering angle MA generated by the turning control unit 48. Therefore, autonomous adaptation is performed by calculating the yaw rate gain CkrRT (see FIG. 30).

  Further, the control coefficient calculation unit 250 uses the lateral deviation Yer calculated by the shape calculation unit 42 and the steering angle MA generated by the turning control unit 48, and the slip angle deviation sYer according to the above equation (105). Is calculated. When the calculated slip angle deviation sYer is larger than the slip angle deviation prescribed value Yebeta, the control coefficient calculation unit 250 calculates the slip angle gain CkbRT according to the equation (107). Further, when the calculated slip angle deviation sYer is smaller than −Yebeta, the control coefficient calculation unit 250 calculates the slip angle gain CkbRT according to the equation (109).

  The control coefficient setting unit 246 uses the yaw rate gain CkrRT and slip angle gain CkbRT calculated by the control coefficient calculation unit 250 to reset the yaw rate gain Ckr and slip angle gain Ckb. Similarly to the control coefficient calculation setting unit 46, the control coefficient setting unit 246 sets the lateral deviation convergence time constant Ty and the correction coefficient Kt according to the following condition of the lane position calculation unit 40.

  In addition, about the other structure and effect | action of the vehicle control apparatus which concern on 4th Embodiment, since it is the same as that of 2nd Embodiment, description is abbreviate | omitted.

  As described above, according to the vehicle control apparatus according to the fourth embodiment, the slip angle gain and the yaw rate gain can be adapted according to changes in the characteristics of the vehicle.

  In the fourth embodiment, adjustment of slip angle gain, autonomous adaptation of yaw rate gain estimation by slip angle estimation, and detection of turning curvature and lateral deviation are performed almost simultaneously. This can be achieved by adjusting the detection time and the like according to the correlation between detection equations for detecting the target course shape as the turning curvature ρ and the lateral deviation Yer from the image data of the road image ahead.

  In the fourth embodiment, the case where CkbRT and CkrRT are detected and reset by the convergence of the errors of the steering angles MaR and MaY and the error of the lateral deviation Yer has been described as an example. However, the present invention is not limited to this. Instead, CkbRT, CkrRT, and other control coefficients may be detected and reset by optimization by convergence of other detection values such as a turning curvature error.

  In the second to fourth embodiments, an upper limit value and a lower limit value are set for each of CkrRT and CkbRT, and an adaptive value is set if the upper limit value and the lower limit value are exceeded. Also good. For example, excessive autonomous adaptation may not be performed in a state where the vehicle behavior is very small (eg, the lateral acceleration is equal to or less than a specified value). In this case, the value of Ckr and Ckb to be reset may be set to 0 or a value according to the initial setting value.

  In the second to fourth embodiments, the case where the detected CkrRT and CkbRT values are reset as the yaw rate gain and the slip angle gain has been described as an example. However, the present invention is not limited to this. Instead, the weighted average with the previous value may be reset as the yaw rate gain and slip angle gain.

  Further, the yaw rate gain may be reset by combining the calculation methods of the yaw rate gain CkrRT described in the second to fourth embodiments. Further, the slip angle gain may be reset by combining the calculation methods of the slip angle gain CkbRT described in the second embodiment and the fourth embodiment.

<Example>
Next, the result of trajectory tracking control using the method described in the fourth embodiment is shown.

  31 and 32 show the time series results of the course follow-up trial calculation when the vehicle weight m is x1.2 (20% increase from the standard, equivalent to five passengers). In the following course, the straight section (S) and steady turning section (R) were connected by the clothoid relaxation section (C) and followed the center of the course. The lateral acceleration of turning was 0.2G, the steady turning was R400, and the vehicle speed was 100km / h.

  FIG. 31 shows the results when the specified values are used for Ckr and Ckb. There is an increase in steady lateral deviation when turning.

  In FIG. 32, CkrRT (true value: solid line, calculated value: two-dot chain line) and CkbRT (true value: broken line, calculated value: one-dot chain line) are calculated by the method described in the fourth embodiment, and the number of vehicles increases. The time series results of the course follow-up trial calculation when adapted to the state are shown. It can be seen that the increase in steady lateral deviation during turning is reduced due to adaptation.

  Next, the result of trajectory tracking control using the method described in the third embodiment is shown. The vehicle speed at which the turning lateral acceleration is 0.2G is set according to the turning radius.

  FIG. 33 shows the lateral deviation rms value of the tracking error in the course tracking trial calculation when the vehicle weight m is set to x1.0 (reference state, equivalent to one passenger), and x1.2 (increased by 20%, equivalent to five passengers). The follow-up conditions were the same as those shown in FIGS. 31 and 32, and a turning radius assuming normal traveling was set. The travel distance varies depending on the turning radius.

  When the specified value is used, a high-speed (large turning radius) follow-up error increases due to a set value error, and an increase in error becomes remarkable as the vehicle weight increases. The tracking error is reduced by the adaptation (CkrRT) using the method described in the third embodiment, and it is possible to confirm improvement in tracking accuracy and stability.

  In the above embodiment, the case where the lane mark is extracted from the image captured by the front camera has been described as an example. However, the present invention is not limited to this, and the rear camera is further used. The lane mark may be extracted from the image captured by the above.

  In addition, the case where the reference direction is the camera imaging direction and the angle notation of the two points on the target course is expressed by the depression angle and the azimuth angle with respect to the camera imaging direction from the camera viewpoint has been described as an example. is not. The angle notation of two points on the target course may be expressed by a depression angle and an azimuth angle with respect to the reference direction from the camera viewpoint using a reference direction other than the camera imaging direction.

  Moreover, although the case where the depression angle and the azimuth angle are acquired for two points on the target course has been described as an example, the present invention is not limited to this, and the depression angle and the azimuth angle are determined using the two points on the lane mark as detection points. You may make it acquire.

  Moreover, you may apply to mobile bodies other than a vehicle, if it is a mobile body which changes a movement position by turning motion.

DESCRIPTION OF SYMBOLS 10 Vehicle control apparatus 12 Front camera 14 Vehicle speed sensor 16 Yaw rate sensor 18 Steering mechanism 20 Computer 22 Image acquisition part 24 Lane mark extraction part 26 Target course detection part 28 Vehicle information acquisition part 30 Detection coefficient calculation part 32 Detection point reference value calculation part 34 Detection point acquisition unit 36 Shape coefficient calculation unit 38 Detection coefficient setting unit 40 Lane position calculation unit 42 Shape calculation unit 44 Vehicle attitude calculation unit 46 Control coefficient calculation setting unit 48 Turn control unit 210 Vehicle control device 220 Computer 246 Control coefficient setting unit 250 Control coefficient calculator

Claims (8)

A moving body control device for controlling a turning movement of a moving body that changes a moving position by a turning movement,
Speed detecting means for detecting the moving speed of the moving body;
Course detection means for detecting a lane mark or a target course from an image picked up by an image pickup means for picking up the front or rear of the moving body,
Angle detection means for detecting a depression angle and an azimuth of the direction of the point with respect to a reference direction for each of a plurality of points on the lane mark or target course detected by the course detection means,
Based on the depression angle and the azimuth angle detected for each of the plurality of points by the angle detection means, the turning curvature of the lane mark or the target course and the lateral deviation of the moving body with respect to the lane mark or the target course are calculated. Shape calculation means;
Based on the turning curvature of the lane mark or target course calculated by the shape calculating means, the lateral deviation of the moving body with respect to the lane mark or target course, and the moving speed detected by the speed detecting means, Turning control means for controlling the turning motion of the moving body so as to follow a target course;
A moving body control apparatus including: For each of the plurality of points, a curvature coefficient, which is an azimuth coefficient per curvature, is calculated based on the depression angle of the point and the viewpoint height of the imaging unit, and the depression angle of the point is calculated for each of the plurality of points. A shape factor calculating means for calculating a lateral deviation coefficient that is an azimuth coefficient per lateral deviation based on the azimuth angle and the viewpoint height of the imaging means,
The shape calculation means includes a depression angle and an azimuth angle detected for each of the plurality of points by the angle detection means, and the curvature coefficient and the lateral deviation calculated for each of the plurality of points by the shape coefficient calculation means. The mobile body control device according to claim 1, wherein a turning curvature of the lane mark or the target course and a lateral deviation of the mobile body with respect to the lane mark or the target course are calculated based on a coefficient.   The turning control means realizes a turning yaw rate that converges a lateral deviation of the moving body with respect to the lane mark or the target course according to the turning curvature of the lane mark or the target course calculated by the shape calculating means. The moving body control device according to claim 1, wherein the moving body controls a turning motion of the moving body.   The turning control means realizes a turning yaw rate that converges a lateral deviation of the moving body with respect to the lane mark or the target course according to the turning curvature of the lane mark or the target course calculated by the shape calculating means. The mobile body control device according to claim 3, wherein a steering angle of the mobile body is controlled by calculating a steering angle of the mobile body. For each of the plurality of points, a curvature coefficient, which is an azimuth coefficient per curvature, is calculated based on the depression angle of the point and the viewpoint height of the imaging unit, and the depression angle of the point is calculated for each of the plurality of points. A shape factor calculating means for calculating a lateral deviation coefficient that is an azimuth coefficient per lateral deviation based on the azimuth angle and the viewpoint height of the imaging means;
Yaw rate detection means for detecting the yaw rate of the moving body;
Based on the yaw rate of the moving body detected by the yaw rate detecting means, the moving speed detected by the speed detecting means, and the lateral deviation coefficient calculated for each of the plurality of points by the shape coefficient calculating means. , Calculate the slip angle of the moving body,
The moving body with respect to the steering angle based on the yaw rate of the moving body detected by the yaw rate detecting means, the calculated slip angle of the moving body, and the steering angle calculated by the turning control means. 5. The moving body control device according to claim 4, further comprising control gain calculating means for calculating a slip angle gain of the moving body with respect to the yaw rate gain and the steering angle.   The steering based on the turning curvature of the lane mark or target course calculated by the shape calculating means, the moving speed detected by the speed detecting means, and the steering angle calculated by the turning control means. The mobile body control device according to claim 4, further comprising control gain calculation means for calculating a yaw rate gain of the mobile body with respect to a corner.   The said control gain calculation means calculates the slip angle gain of the said mobile body with respect to the said steering angle according to the lateral deviation of the said mobile body with respect to the said lane mark or target course calculated by the said shape calculation means. Mobile control device. Further comprising posture calculating means for calculating the posture of the moving body based on the slip angle calculated using the slip angle gain of the moving body calculated by the control gain calculating means;
The turning control means calculates the steering angle based on the yaw rate gain of the moving object calculated by the control gain calculating means, and controls the steering angle of the moving object;
The shape calculation unit is configured to determine the lane mark or the target based on the depression angle and the azimuth angle detected for each of the plurality of points by the angle detection unit and the posture of the moving body calculated by the posture calculation unit. The moving body control device according to claim 5 or 7, wherein a turning curvature of a course and a lateral deviation of the moving body with respect to the lane mark or the target course are calculated.