JP2748714B2 - Steering control device using multiple regression equation for autonomous vehicle - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a steering control based on a multiple regression equation in an autonomous vehicle, that is, a vehicle that automatically travels according to a preset route.

[0002]

2. Description of the Related Art As a conventional steering control device for an autonomous traveling vehicle, for example, a video camera installed at the front of the vehicle captures an image of a scene in front of the vehicle, and a sign indicating a traveling road in the captured image, that is, a roadside or center. There is a device that detects a white line indicating a line and controls a steering angle so as to travel along the white line. FIG. 11 is a plan view for explaining a control method in the steering control device as described above. FIG.
, 10 is a vehicle, and 11 is a white line indicating a road edge. In this control method, a scene in front of the vehicle 10 is imaged by an imaging device such as a video camera, and data (d ₁ , y ₁ , t ₁ ), (d ₂ ,
y ₂ , t ₂ ),... are used to calculate the equation y = _fr (x) of the white line 11 in the xy coordinates with the vehicle center 0 as the origin. Then, distance data d ₀ and tangent angle data t ₀ from the vehicle to the white line on the side are estimated and calculated based on the equations, and the steering control is performed using these values. Here, y is the distance to a predetermined point in front of the vehicle, d is the distance from the predetermined point to a white line perpendicular to the vehicle traveling direction, and t is the tangent at the point where the line perpendicular to the vehicle traveling direction intersects the white line. Is the corner. The tangent angle is an angle between the traveling direction of the vehicle and the tangent of the curve.

[0003]

In the conventional steering control as described above, a curve equation of a white line is approximately estimated based on information on a white line ahead, and the vehicle position from the white line immediately beside is calculated using the equation. Is calculated. But,
In the conventional estimation method, there is a problem that an error occurs between the actual white line and the calculated white line position, and the error affects the steering control, and it is difficult to smoothly travel along the white line. .

The present invention has been made to solve the above-mentioned problems of the prior art, and has as its object to provide a steering control device capable of performing smooth automatic steering adapted to human senses. And

[0005]

Means for Solving the Problems In order to achieve the above object, the present invention is configured as described in the claims. That is, in the present invention, the distance information to the sign obtained from the imaging means such as a video camera,
That is, distance information from a point ahead of the vehicle L (m) to a sign (for example, a white line indicating a road edge or a center line) indicating a traveling direction perpendicular to the vehicle traveling direction is input, and the steering angle to be steered at that time is input. Is output by multiple regression analysis, and the present and past information is used as the white line information. Regarding multiple regression analysis, "" Analysis of multivariate data in industry ""
Okuno Chuichi, published by the Japan Science and Technology Federation.

[0006]

As shown in FIG. 4, the change in the steering angle when a human is maneuvering and traveling on a curved road is L
(M) The characteristic changes with a similar characteristic after a predetermined time delay from the change in the distance from the preceding point to the white line in the direction perpendicular to the vehicle traveling direction. Therefore, by detecting the distance to the white line, using the current distance and the past distance, and calculating the steering angle by the constant of the multiple regression equation, it is possible to perform automatic steering very similar to that performed by a human being. I can do it. The above constant may be determined by running a predetermined curved road in advance and determining the constant at that time, and setting the curvature of the traveling road to a range including the above curvature. In this case, using the multiple regression equation can absorb the error even if the curvature is slightly different.

It is also possible to determine the above constants for traveling roads of various curvatures, detect the curvature of the traveling road from image data, and select a constant according to the curvature at that time. . In addition, the above constant can be calculated by learning during traveling.

[0008]

FIG. 1 is a block diagram of an embodiment of the present invention, and FIG. 2 is a side view of an autonomous vehicle showing a mounting position of a video camera. As shown in FIG. 2, a video camera 2 is attached to a front end of the autonomous vehicle 1. A white line 3 serving as a sign indicating the traveling path is drawn at both ends of the traveling path. The white line 3 is detected by the video camera 2 and the steering control is performed as described in detail later. It automatically travels on the route that has been set.

The apparatus shown in FIG. 1 comprises an imaging means (video camera) 201, an image processing means 202, a map information means 203, a traveling control means 204, a steering control means 205, and a steering driving means 206. As shown in FIG. 2, the video camera 201 is attached to the front end of the traveling vehicle, captures an image of a scene ahead of the vehicle, converts the image into an electric signal, and outputs the electric signal. The image processing means 202 detects a distance from a point a predetermined distance L (m) before the vehicle to a white line in a direction perpendicular to the vehicle traveling direction based on the image captured by the video camera 201. The map information means 203 stores information such as map data on a route to be traveled. This map information includes, for example, information such as nodes and paths on the traveling road. Note that a node is a point at a boundary between a straight part and a curved part of a road, and a path is a distance between a certain node and an adjacent node.

[0010] The traveling control means 204 is provided with the map information means 20.
Based on the information stored in 3, route information according to the designated route is output. As this route information, for example,
There are information such as a traveling vehicle speed, a traveling distance, a traveling route (where to turn at a branch point), and a turning radius of a curved portion. The steering control unit 205 determines the steering angle when traveling according to a predetermined route based on the route information given from the traveling control unit 204 and the white line information given from the image processing unit 202. The calculation is performed by performing a multiple regression analysis which is a feature of the present invention. The contents of this calculation will be described later.
The steering driving means 206 drives an actuator according to the steering angle obtained by the steering control means 205 to drive the steering device. Note that the above image processing means 2
02, the map information means 203, the traveling control means 204, and the steering control means 205 can be configured using, for example, a microcomputer.

Next, the operation will be described. The present invention is characterized by a steering angle calculation by multiple regression analysis in the steering control means 205, and that point will be described in detail. FIG.
FIG. 3 is a plan view for explaining the operation of the present embodiment. FIG.
As shown in, when the vehicle is traveling on a straight path through the curved road from the straight path, v vehicle center 0 at the position of _{0 0}
The distance from the point L (m) ahead to the white line in the direction perpendicular to the vehicle traveling direction is x ₀ , and the steering angle at that time is S ₀ . In the above case, for example, the steering angle S (deg) and the elapsed time t (se) when the human driver operates the steering device to drive the vehicle from the * Start position to the * End position
The relationship with c) is as shown in FIG. Also, L
(M) the elapsed time and the change of the white line distance information x _m in destination t
Is as shown in FIG.

As can be seen by comparing FIGS. 4 (a) and 4 (b), in the curved part, the steering angle change (b) is similar to the change (a) in the white line distance information change (a) in front of the vehicle. Appears as. Therefore, in FIG. 4B, an appropriate value can be obtained by estimating the steering angle S ₀ after t ₀ seconds by the change of the white line information at that time. Further, as the change value of the white line information, the change value of x value per one control cycle (one time differential value),
Further, the change value of the value per control cycle (two-time differential value)
Was used. In the multiple regression equation, this information can be defined innumerably, but as a result of the experiment, the present and past (1) and the past (2)
Has been found to be sufficient. Therefore, the current white line distance information x ₀ , the value x ₋₁ one control cycle before, and the value x ₋₂ two control cycles before are used as input values. In this case, a relational expression indicating the steering angle S ₀ after t ₀ seconds (regression function f)
S ₀ = f (x ₀ , x ₋₁ , x ₋₂ ) = α · x ₀ + β · x ₋₁ + γ · x ₋₂ + δ (Equation 1) Using the above (Equation 1) and all S, x data from the Start position to the End position, α,
β, γ, and δ are estimated.

FIG. 5 is a graph showing actually measured values of the relationship between the steering angle S and the elapsed time t. The driver can drive the vehicle along a road including a curved portion (r = 15 m) as shown in FIG. 3 at a speed of 10 km / h. Is an example in which the vehicle is operated by operating the steering device, and the regression function formula (f) is estimated based on the actual steering angle at that time and the data of the white line information 5 m ahead obtained by the imaging device. In FIG. 5, a solid line indicates a value when the driver operates the steering device to drive, a broken line indicates 5 m visual information, and a dashed line indicates a regression function. In this case, S = −0.84581 · x ₀ −7.13965 · x ₋₁ + 0.21617 · x ₋₂ +15.4842, as shown by the one-dot chain line in FIG. It can be seen that the agreement is approximately good.

FIG. 6 shows an example in which steering control is actually performed using the above equation. In FIG. 6, the visual sense is a characteristic of the present embodiment, and the ultrasonic wave measures, for example, a distance in a lateral direction from a guardrail installed along a traveling path to the vehicle with an ultrasonic sensor, and the like. This is a characteristic when the steering angle is P (proportional), I (integral), and D (differential) controlled using the value. As shown in FIG. 6, it can be seen that the control according to the present embodiment performs the steering control clearly and smoothly. In addition, since the steering operation is controlled by a control formula obtained based on data obtained when a human operates the steering, it is possible to reproduce control close to a human sense. FIG. 7 is a flowchart showing the calculation processing in the above embodiment.

Next, another embodiment of the present invention will be described. In this embodiment, the coefficients α, β, γ,... Δ are estimated while traveling by learning. As shown in FIG. 8, white line distance data x (n) ahead of L (m) (where n
When the output rudder angle S (n) is determined using (current), first, as shown in FIG. 9 (a), forward L ₁ , L ₂ (m)
Is calculated from the white line information (x ₁ , t ₁ ) _L1 and (x ₂ , t ₂ ) _{L2 at} the n-th order equation y = L (x). However, n is, for example, n = 3.

[0016] Then with ^{x = L ~ 1 (y)} ( described later in detail), using the value X _L ~ ¹ (0) when the y = 0, this is an evaluation value of the vehicle movement. That is, when the vehicle is traveling at a position from the white line for example 2m ^{_{is, 2.0 ≒ | X L ~ 1}} (0) |
The output steering angle S (n) ′ which should be,
^{_{Is, 2.0> | X L ~ 1}} (0) | S (n) when the '= S (n) + δ 2.0 ≦ | X L ~ 1 (0) | when the -S (n) '= S (n) -δ. However, S (n) is positive in the clockwise direction and negative in the counterclockwise direction. In the case of following the left white line, δ ≧
0, δ <0 in the case of following the right white line. In other words, as data By updating the old S (0) data among the N + 1 data of S (N) and inserting the above data, the α, β, γ,... The steering angle is calculated and output based on the calculated steering angle. By controlling as described above, the coefficient of the control formula is updated while traveling, and stable traveling control can be performed after traveling to some extent.

Here, x = L ＝ ¹ (y) will be described. FIG. 9B is an image of the front of the vehicle taken by the video camera. A white line in front is visible but a white line right beside the vehicle is not visible. Therefore, an equation of y = L (x) is established, and _XL - ¹ (0) is estimated.

Next, FIG. 10 is a flow chart showing the operation contents of still another embodiment of the present invention. As shown in FIG. 10, the current vehicle speed v (km / h) is classified according to the _magnitude , and the multiple regression coefficients α, β, γ,... _{Are determined} for each of the classified _VcLass. Accordingly, once the vehicle automatically travels from a low speed to a high speed, stable steering control can be performed at all speeds thereafter. The above classification is for example a _{0 ≦ v <10 V cLass (} v) = V S when the, and ₁₀ ≦ v <20 V cLass when the _{(v) = V M, 20} ≦ v <30 When is V
_{cLass (v)} = to as V _B.

[0019]

As described above, according to the present invention, a relational expression in which the distance information of the front white line obtained by the imaging device is input and the steering angle at that time is output is estimated by multiple regression analysis, and Since the white line information is configured to use the current and past information, the multiple regression equation is estimated based on the imaging device-steering data when a person steers the vehicle. Control is performed smoothly with characteristics close to human feeling. In addition, compared to the steering control based on information calculated geometrically as in the conventional example, it is possible to follow white lines such as cycloid curves regardless of straight lines or curves, and smooth steering control is possible. Is obtained.

[Brief description of the drawings]

FIG. 1 is a block diagram of a first embodiment of the present invention.

FIG. 2 is a side view of the vehicle showing a mounting position of the video camera.

FIG. 3 is a plan view for explaining the operation of the first embodiment.

4 (a) is the elapsed time and the change of the white line distance information x _m t
FIG. 4B is a characteristic diagram showing a relationship between the steering angle S (deg) and an elapsed time t (sec).

FIG. 5 is a characteristic diagram showing actually measured values of a relationship between a steering angle S and an elapsed time t.

FIG. 6 is a characteristic diagram when steering control is actually performed.

FIG. 7 is a flowchart illustrating a calculation process according to the first embodiment.

FIG. 8 is a plan view for explaining the operation of the second embodiment of the present invention.

FIG. 9 is a plan view for explaining the operation of the second embodiment.

FIG. 10 is a flowchart illustrating a calculation process according to a third embodiment of the present invention.

FIG. 11 is a plan view for explaining a calculation method in a conventional steering control device.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Vehicle 2 ... Video camera 3 ... White line 10 ... Vehicle 11 ... White line 201 ... Image pickup means (video camera) 202 ... Image processing means 203 ... Map information means 204 ... Travel control means 205 ... Steering control means 206 ... Steering drive means

Continuation of the front page (58) Field surveyed (Int.Cl. ⁶ , DB name) G05D 1/02 JICST file (JOIS)

Claims (1)

(57) [Claims] 1. An image pickup apparatus, comprising: an image processing means; an image processing means; a map information means; a traveling control means; a steering control means and a steering driving means; An image of a scene in front of the vehicle is captured, converted into an electric signal and output, and the image processing means is configured to move in a vehicle traveling direction from a point a predetermined distance before the vehicle based on an image captured by the imaging device. The map information means is for detecting a distance to a sign indicating a traveling path in a right angle direction, wherein the map information means stores information relating to map data of the travel path, and the travel control means is provided for the map information means. Output the route information according to the designated route based on the information stored in the steering control unit, wherein the steering control unit receives the route information given from the traveling control unit and the route information given from the image processing unit. A relational expression that, based on the distance information to the intellect, the steering angle when traveling according to a predetermined route determined in advance, the distance information to the sign is input, and the steering angle to be steered at that time is output. Is calculated by estimating by means of multiple regression analysis, and uses the current information and past information as the distance information to the sign, the steering drive means,
A multiple regression-type steering control device for an autonomous vehicle, which drives a steering device according to a steering angle obtained by the steering control means.