JPH05216530A - Automatic driving controller for vehicle - Google Patents

Abstract

PURPOSE:To accurately run a vehicle on a target course even in the case of the difference between the running condition (the wet state on the road surface or the like) for teaching operation and that for automatic running operation. CONSTITUTION:At the time of teaching operation, a vehicle position (x), a vehicle velocity v, a transverse deviation DELTA1, a yaw angle DELTAtheta, a yaw rate omega, and an extent F of steering are stored in maps 141 and 143. The radius of turning is detected by the vehicle velocity V and the yaw rate omega, and correspondence relations between the extent of steering and the radius of turning (correspondence relations between the extent of steering and the curvature of turning) are calculated as an ideal model. At the time of actual automatic running operation, the taught extent F of steering is corrected by a forecasted error DELTAl-L.DELTAtheta (DELTAl is the transverse deviation error and DELTAtheta is the yaw angle error) at the position Lm ahead to obtain an extent delta of steering, and a gain K is so adjusted by an adjuster 144 that an error (e) between a curvature rho0 of turning and an actual curvature rho of turning in the ideal model of this extent delta of steering is dissolved, and it is outputted.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an automatic steering control device for a vehicle, and more particularly to the control contents of the automatic steering control device for automatically traveling a vehicle according to a predetermined guideway.

[0002]

2. Description of the Related Art Autonomous vehicles that automatically travel on a predetermined running path are well known and are used in automated guided vehicles in factories and in durability running tests. Such a self-driving vehicle has, for example, an induction cable provided on the track, and while operating the steering wheel, accelerator, brake, etc. with an actuator while detecting the position of the induction cable with the magnetic field sensor provided on the vehicle, It realizes automatic driving, which allows the vehicle to run faithfully on the course.

Further, in the past, Japanese Patent Laid-Open No. 63-33
As disclosed in Japanese Patent Laid-Open No. 15-15, there is a system in which an actual course is run by manned operation to collect course information, and the vehicle is provided with information necessary for setting a traveling course in advance. According to this, even if the terrain is relatively complicated, it is possible to smoothly travel without deviating from the predetermined course.

However, in such a conventional automatic vehicle control, since the vehicle is controlled by a program created mechanically, it is not possible to carry out the fine-tuned driving that a human actually controls, and the ride comfort is improved. There was a problem of being bad.

Generally, in an autopilot, running the course faithfully is the most important issue. Therefore, since high-speed control with a high feedback gain is performed, it is very different from the one operated by a human, and the control There was a problem that stability deteriorated. Therefore, the applicant of the present application has previously proposed the following automatic vehicle control device in Japanese Patent Application Laid-Open No. 2-231609. That is, first, the driver actually travels on the taxiway to teach the taxiway as the target course, and the operation amounts of various actuators during the teaching operation, for example, the steering amount are detected and stored in the storage means together with the target course. .. Then, when the vehicle is controlled to travel in accordance with the teaching content, the prediction error when traveling from the current position to a predetermined distance ahead is sequentially calculated, and the teaching content is corrected according to the prediction error to calculate the operation amount of the actuator. Is.

[0006]

However, the running condition when the driver actually runs and teaches the target course and the running condition when the vehicle automatically runs according to the teaching content according to the teaching contents are the friction coefficient of the road surface. There was a problem that it was not always possible to drive accurately on the target course when the value varied depending on μ and changes in vehicle speed.

That is, when the friction coefficient μ of the road surface and the vehicle speed are different from the values at the time of teaching due to the wetness of the road surface, the front and rear wheel cornering forces and the motion characteristics of the vehicle itself change.
Since the steering characteristic of the vehicle is changed, the turning radius of the vehicle is different from that at the time of teaching even though the vehicle is controlled with the same steering amount as the teaching content, and a course deviation occurs.

Of course, it is considered that the teaching data is made to match the running condition by using such a difference in the running condition and using a prediction error caused by the difference in the running condition. No. 3-13595), the conventional feedback control contents are forced to be changed.

The present invention has been made in view of the above problems of the prior art, and its object is to provide teaching data and conventional feedback control contents even when the traveling conditions are different between the teaching operation and the actual automatic pilot traveling. An object of the present invention is to provide an automatic steering control device for a vehicle, which is capable of accurately traveling on a target course without changing.

[0010]

In order to achieve the above object, an automatic vehicle control system for a vehicle according to the present invention is a target course in which a driver drives and teaches a predetermined taxiway,
And teaching data storage means for storing the steering amount and turning radius when driving this target course as teaching data, and a prediction error calculation means for calculating a prediction error when the vehicle travels a predetermined distance from the current position, Feedback control means for correcting the teaching steering amount so as to eliminate this prediction error, and model calculation means for calculating the correspondence between the steering amount and the turning radius during teaching operation as an ideal model,
Model error calculating means for calculating an error between a turning radius in this ideal model and a turning radius during automatic driving, and an actuator operation for correcting the teaching steering amount so as to eliminate the model error and calculating the operation amount of the actuator. It is characterized by having an amount adjusting means.

[0011]

As described above, when the driving condition during teaching driving differs from that during actual autopilot driving, the front / rear wheel cornering force and the motion characteristics of the vehicle itself change, and the steering characteristics of the vehicle change. Even if the vehicle is controlled with the same steering amount as the teaching content, the turning radius of the vehicle is different from that at the time of teaching.

Therefore, in the present invention, the ideal correspondence between the two data is calculated from the steering amount and the turning radius during the teaching operation, and even when the teaching operation and the traveling conditions change during the actual automatic pilot traveling. The final actuator operation amount is always adjusted so as to maintain this ideal correspondence.

Then, the turning radius during the teaching operation matches the turning radius during the actual automatic steering traveling (in this case, the steering amount during the teaching driving and the steering amount during the automatic steering traveling change in traveling conditions). It will be possible to drive accurately on the target course.

Note that, by adopting a configuration in which an ideal model is extracted from the steering amount-turning radius relationship during teaching operation and the steering amount is corrected so as to match the ideal model, prediction error and teaching data are obtained. Does not need to be individually adjusted according to the difference in traveling conditions, and the purpose can be achieved without changing the conventional control content.

[0015]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of an automatic vehicle control system according to the present invention will be described below with reference to the drawings.

FIG. 2 shows the configuration of this embodiment.
Front magnetic field sensor 110 for detecting lateral deviation of a vehicle
a, front lateral displacement detection circuit 111a, rear magnetic field sensor 11
0b, a rear lateral deviation detection circuit 111b is provided, and the lateral deviation of the vehicle is obtained by detecting the magnetic field output from the induction cable 20 provided in the guide path, and the yaw angle is obtained. That is, the front lateral displacement is Yf, and the rear lateral displacement is Yf.
Letting r be the lateral displacement Y of the vehicle 28, Y = (Yf + Yr) / 2, and letting the total length of the vehicle 28 be a, the yaw angle θ of the vehicle is θ = (Yf−Yr) / a. ..

A magnet pickup 114 for determining the vehicle speed is provided, and the vehicle speed is detected by detecting the number of rotations of the magnet rotating plate 113.

Then, the antenna 115 for detecting the position of the vehicle on the taxiway and the reference position detection circuit 116.
Is provided, and the radio wave output from the position beacon provided at each predetermined point of the taxiway is received by the antenna 115 and supplied to the reference position detection circuit 116, whereby the reference position on the course is detected.

Further, an arithmetic processing circuit ECU 24 for inputting detection signals from the above-mentioned various sensors to perform arithmetic processing, and a feedback processing circuit 26 for feeding back operation amounts of various actuators for steering the vehicle to perform predetermined processing.
Is provided to control the steering 30 of the vehicle.

In the present embodiment, the manual switch 3 for switching between the teaching operation mode and the automatic operation mode.
4 and an indicator lamp 36 for identifying this mode state
Is provided near the driver's seat of the vehicle.

FIG. 1 shows a principle explanatory diagram of a calculation process for obtaining a final actuator operation amount from teaching data. When the current position of the vehicle is away from the target course by Δl and the yaw angle is Δθ, the predetermined distance Lm
An error of L.DELTA..theta. Is newly generated in the above, and therefore, the prediction error of Lm ahead is .DELTA.l-L.Δθ. In order to eliminate this prediction error, a predetermined gain K is added to this prediction error.
_{The p} is constantly added to the teaching data F given by the teaching operation and corrected. The steering amount δ is usually used as the teaching data. When the driving condition during teaching operation is the same as the driving condition during automatic pilot control, it is possible to drive on the target course by this control. It is impossible to drive on the target course only with such predictive error control. This is because the turning radius is different even if the steering amount is the same, under different traveling conditions. Therefore, in this embodiment, an ideal model is calculated from teaching data during teaching operation, and steering amount control is performed so as to match the output of this ideal model. As the ideal model, the relationship between the steering amount and the turning radius is adopted,
During automatic driving, the ideal turning radius corresponding to the steering amount at that point is output as the output from this ideal model. Then, this ideal turning radius is compared with the actual turning radius during automatic driving, and the steering amount data is corrected by gain adjustment by the adjuster so as to eliminate the error and supplied to the actuator.

FIG. 6 schematically shows the relationship between the steering amount δ and the turning curvature ρ which is the reciprocal of the turning radius. The one-dot chain line in the figure shows an ideal correspondence obtained from the steering amount and turning radius data during teaching operation, and the solid line in the figure shows the correspondence obtained from the steering amount and turning radius during actual automatic pilot traveling. Note that, in reality, these correspondence relationships are not linear, but in FIG. 6, it is assumed that they are linear relationships for convenience of explanation.

Here, when the traveling conditions differ between the teaching operation and the actual automatic steering traveling, the turning radius differs even with the same steering amount, and therefore the one-dot chain line and the solid line do not match as shown in the figure. Therefore, in actual traveling, the turning radius obtained by steering with the same steering amount δ _{0 as} in teaching operation is ρ
It becomes ρ ₁ instead of ₀ . Therefore, the steering amount is set to δ ₀
Therefore, if the correction is made to δ ₁ , the turning radius ρ ₀ can be obtained even under actual traveling conditions, and the vehicle can travel on the target course.

FIG. 3 shows a more detailed block diagram of the arithmetic processing of this embodiment, and FIG. 4 shows a processing flow chart of this embodiment. First, in FIG. 4, it is determined whether or not the teaching operation is completed (S101). When the teaching operation is in progress, the traveling distance x, the steering amount δ, the vehicle speed V,
The yaw rate ω, the lateral deviation Δl, the yaw angle Δθ, and the turning curvature ρ that is the reciprocal of the turning radius are detected and stored (S103).
~ S104). Specifically, the position x detected from the position beacon 22 or the like during teaching operation and the lateral deviation Δl, yaw angle Δθ, and yaw rate ω at this position x are averaged by the averaging unit 241.
Are averaged and stored in the target course map 141.
Further, the steering amount δ at which the driver operates the steering wheel during the teaching operation is similarly averaged by the averaging unit 242 and stored in the steering amount map 142. Therefore, the vehicle speed V, the yaw rate ω, the turning curvature ρ (= ω / V), and the steering amount δ as a function of the position x are stored in the teaching map as time series data, as shown in S104. ..

When the teaching operation is completed, ideal model calculation is performed (S102). This ideal model calculation is based on the time series steering amount data δ ₀ stored in the teaching map,
δ _1, δ ₂ ...... chronological turning curvature data [rho ₀ corresponding to the steering amount data of Toko, [rho _1, it is calculated as the corresponding relationship between the [rho ₂ .......

Hereinafter, this ideal model (ARMA model)
A method for obtaining is described. First, the ideal model is described by the following mathematical formula.

Ρ (k) + a ₁ ρ (k-1) + a ₂ ρ (k-2) = b ₁ δ (k-1) + b ₂ δ (k-2) ρ (k) = ω (k) / V (k) where k is a parameter indicating the traveling position. Now, as teaching data, δ (1), δ (2), δ (3), ..., δ
(N-1) and ρ (1), ρ (2), ρ (3), ..., ρ
It is assumed that (N) is detected and stored. An ideal model can be obtained by determining a ₁ , a ₂ , b ₁ , and b ₂ in the above equation based on these data. Substituting the teaching data into the above equation, ρ (3) = − a ₁ ρ (2) −a ₂ ρ (1) + b ₁ δ (2) + b ₂ δ (1) ρ (4) = − a ₁ ρ ( 3) -a ₂ ρ (2) + b ₁ δ (3) + b ₂ δ (2) ... ρ (N) =-a ₁ ρ (N-1) -a ₂ ρ (N-2) + b ₁ δ ( N−1 recurrence formulas of N−1) + b ₂ δ (N−2) are obtained. Then, from these N−2 recurrence formulas, a ₁ , a
_It suffices to determine ₂ , b ₁ and b ₂ .

Expressing the N-2 recurrence formulas by a matrix,

[Equation 1] When this is expressed as ρ _N = Φ _N a, the estimated value a _{N of} a is calculated by the least-squares method: a _N = (Φ _N ^T Φ _N ) ⁻¹ Φ _N ^T ρ _N , and a ₁ , A ₂ , b ₁ , b ₂ can be obtained.

The above-mentioned method has shown the case where a ₁ , a ₂ , b ₁ and b ₂ are determined from N teaching data.
On the other hand, when the N + 1th teaching data is newly added, it is possible to sequentially obtain the N + 1th teaching data and the result a _N obtained from the N teaching data.

That is, the N + 1th teaching data is ρ (N + 1) =-a ₁ ρ (N) -a ₂ ρ (N-1) + b
_{Since 1} δ (N) + b ₂ δ (N-1),

[Equation 2]

[Equation 3]

[Equation 4] Holds. Therefore, the following equations are obtained from these equations, and it is possible to obtain them sequentially.

A _{N + 1} = S _{N + 1} −Φ _{N + 1} ^T y _{N + 1} S _{N + 1} = S _N −S _N Φ _{N + 1} Φ _{N + 1} ^T S _N / (1 + Φ _{N + 1} ^T S _N Φ _{N + 1} )

[Equation 5] After the ideal model is calculated in this way, control is performed so as to eliminate the error between the turning curvature ρ ₀ that is the output of the ideal model and the turning curvature ρ during actual automatic pilot traveling. That is, the ideal model calculated by the method described above is stored in the ideal model map 143 (specifically,
Coefficients a ₁ , a ₂ , b ₁ and b ₂ are stored), and an ideal turning curvature ρ ₀ corresponding to the teaching data δ is output, and an error from the actual turning curvature (= ω / V) at this time is output. e = ρ ₀ −ρ
Is calculated (S105). Then, the steering angle δ is adjusted so as to eliminate this error, that is, to match the traveling characteristics during automatic steering traveling with the ideal turning curvature (turning radius).

Specifically, the adjuster 144 is based on the error e.
Is based on the input steering amount δ and the error e

[Equation 6] The adjustment gain K is adjusted by. Then, the teaching data F (steering amount) is read from the steering amount map, and the prediction error amount, that is, the feedback correction amount Δe = Δl−L is read into this F.
The steering angle data δ obtained by adding Δθ is adjusted by the adjustment gain K and output to the actuator (S10
7, S108).

Then, the output steering angle at this time has a value at which the ideal turning radius can be obtained, so that the vehicle can travel correctly on the target course.

As described above, in this embodiment, the ideal model is extracted from the correspondence between the steering angle and the turning radius obtained during the teaching operation, and the steering amount is corrected so as to match the output of the ideal model. As a result, it becomes possible to travel accurately on the target course by using the conventional prediction error control.

In this embodiment, the ideal model was calculated after the teaching operation was completed, but it is of course possible to calculate the model in real time at each point during the teaching operation.

[0036]

As described above, according to the vehicular autopilot control device of the present invention, the ideal model is extracted even when the driving condition during teaching driving and the driving condition during automatic driving are different. Since the teaching steering amount data is corrected so as to match the output of this ideal model, it is possible to travel accurately on the target course.

[Brief description of drawings]

FIG. 1 is a diagram illustrating the principle of an embodiment of the present invention.

FIG. 2 is a configuration block diagram of the embodiment.

FIG. 3 is a block diagram of a calculation processing configuration of the embodiment.

FIG. 4 is a processing flowchart diagram of the embodiment.

FIG. 5 is a graph showing a relationship between a position x, a steering angle and a turning curvature in the embodiment.

FIG. 6 is a graph diagram schematically showing a correspondence relationship between a steering amount and a turning curvature in the embodiment.

[Explanation of symbols]

 20 guidance cable 22 position beacon 24 arithmetic processing circuit ECU 26 feedback processing circuit 141 target course map 142 steering amount map 143 ideal model map

Claims (1)

[Claims] 1. An autopilot control device for a vehicle, which measures a current position and running condition of a vehicle with respect to a taxiway and controls operation of each actuator for automatic piloting based on the position and running condition, which is predetermined. A target course taught by the driver by driving the taxiway, and teaching data storage means for storing the steering amount and the turning radius when driving the target course as teaching data, and the vehicle travels a predetermined distance from the current position. And a feedback control unit that corrects the teaching steering amount so as to eliminate the prediction error, and the correspondence relationship between the steering amount and the turning radius during teaching operation is provided. Is calculated as an ideal model, and a model error for calculating an error between the turning radius in this ideal model and the turning radius during automatic driving A calculation unit, autopilot control system for a vehicle, comprising a an actuator operation amount adjusting means for calculating an operation amount of the corrected actuator the teachings steering amount so as to eliminate the modeling error.