JP2762784B2 - Data calculation device for autonomous vehicles - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an autonomous vehicle that travels unmanned along a reference line such as a white line (center line) provided on a road surface, and more particularly to steering of an autonomous vehicle. For calculating travel data of the vehicle.

[0002]

2. Description of the Related Art As a method for driving an unmanned autonomously traveling vehicle along a predetermined route, a cable, a mark tape or the like is attached to the traveling route, and the vehicle is detected while detecting these with a sensor provided on the vehicle. In addition to the fixed route method in which the vehicle travels, a trackless method in which the front of the vehicle is photographed with an imaging device, a target route (reference line) is extracted from an image of the imaging device, and the vehicle travels along the target route is known. (See, for example, JP-A-64-7110).

In a conventional trackless autonomous vehicle, a reference line provided on a road surface is extracted by image processing as described above, and the distance between a reference point of interest and a reference point d [m] ahead of the vehicle is determined. (Hereinafter, referred to as deviation) y [m], and the steering angle is determined based on the deviation to travel.

In a conventional autonomous vehicle, linear interpolation or least-squares approximation is performed on the deviation data calculated in the previous measurement in order to estimate whether the calculated deviation is accurately measured. In this case, a deviation (linear) curve of a linear function or a quadratic function is obtained, and the deviation to be measured this time is estimated from the deviation curve.

[0005]

However, in the conventional autonomous vehicle, the deviation curve is calculated and the deviation is estimated based only on the data of one image obtained by the previous measurement. However, there is a problem that it is not easy to sufficiently secure the estimation accuracy of the deviation, and in particular, when data is lost during measurement, the estimation accuracy of the deviation is further reduced.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a traveling data computing device for an autonomous traveling vehicle capable of accurately estimating a deviation from a reference line.

[0007]

The present invention is described with reference to FIG. 1 which is a diagram corresponding to the claims. In the present invention, a road surface in front of a vehicle is photographed and a predetermined period is set such that an overlapping portion is formed in an adjacent image. Imaging means a for outputting image data every time; an image processing means b for extracting a reference line extending along the road surface from an image of the imaging means a; and a reference line extracted by the image processing means b. Coordinate calculating means c for calculating the coordinate value, storage means d for storing the coordinate value of the reference line calculated in the past, and a reference line for approximating the coordinate value based on the coordinate value in the storage means d A function calculating means e for calculating a function, and an error between a coordinate value obtained by the coordinate calculating means c and a reference line function obtained by the function calculating means e is obtained. Coordinate values as driving data And a determination means f for outputting the coordinate value of the reference line estimated from the reference line function as travel data if the error is outside the predetermined range. By doing
The above-mentioned object is achieved.

[0008]

The image pickup means (a) takes an image of the road surface in front of the vehicle and outputs image data at predetermined intervals such that an overlapping portion is formed in an adjacent image. The image processing means b includes an imaging means a
The reference line extending along the road surface is extracted from the image of (1), and the coordinate calculation unit c calculates the coordinate value from the reference line extracted by the image processing unit b. The storage unit d stores the coordinate values of the reference line calculated in the past, and the function calculating unit e calculates a reference line function that approximates the coordinate values based on the coordinate values in the storage unit d. Then, the determining means f
The coordinate value obtained by the coordinate calculating means c and the function calculating means e
Is calculated, and if the error is within a predetermined range, the coordinate value obtained by the coordinate calculation means c is output as traveling data, and the coordinate value is stored in the storage means d. Let it. On the other hand, if the error is outside the predetermined range, the coordinate value of the reference line estimated from the reference line function is output as traveling data.

[0009]

Embodiments of the present invention will be described below in detail with reference to the drawings. FIG. 2 is a block diagram showing an autonomous traveling vehicle to which the traveling data calculation device according to the present invention is applied. In FIG. 2, reference numeral 1 denotes an autonomous vehicle (hereinafter, simply referred to as “vehicle”). The vehicle 1 extracts a white line drawn on a road by image processing, and extracts a white line from the vehicle based on the extracted white line. A deviation y [m] at the forward gaze point ahead of d [m] is calculated, a steering angle is determined based on the deviation, and the vehicle runs in accordance with the steering angle. Therefore, although not shown in FIG. 2, the vehicle 1 is provided with a steering angle determining device for determining a steering angle from a deviation, an actuator for steering the vehicle, a speed determining device for determining a vehicle speed, and the like. .

As shown in FIG. 2, the vehicle 1 includes an imaging device 2, an image processing device 3, a storage device 4, a rotary encoder 5, a vehicle position calculating device 6, and a white line estimating device 7.

The image pickup device 2 is, for example, a TV camera, and is fixed in front of the vehicle 1 to photograph a road surface image in front of the vehicle 1. The photographed road surface image data is sent to the image processing device 3.

The image processing device 3 extracts a white line extending along the road surface from the road surface image data sent from the imaging device 2 by a known image processing method, and based on the position of the white line in this image. Te, calculated distance from a predetermined forward gaze point provided (i = 1 to 11 in the present embodiment) forward d _i of the vehicle 1 to the white line, i.e. the deviation y _i. Distance to this forward fixation point (hereinafter, simply referred to as forward fixation point distance) d
_i is stored in the storage device 4 in advance. In the present embodiment, as shown in FIG. 3, the forward fixation point distance d _i and deviation y _i are fixed predetermined reference point in the vehicle 1 (the center of the front wheel tires) to the vehicle 1 to the origin Calculated on a coordinate system (hereinafter, referred to as a vehicle coordinate system). The calculated deviation data is sent to the white line estimation device 7.

The storage device 4 stores the above-mentioned forward fixation point distance d.
_i , and past deviation data described later, weights at the time of function estimation, the position of the vehicle 1 at the previous measurement, specifications of the vehicle 1,
It stores the positional relationship of the reference points provided therein, variables for evaluating deviation data, and the like.

The rotary encoder 5 is attached to the rear wheel tire 8 and sends out the rotation speed of the rear wheel tire 8 to the vehicle position calculating device 6 at every control cycle.

The vehicle position calculating device 6 calculates the rotational speed of the rear wheel tire 8 obtained from the encoder 5, the specifications of the vehicle 1 stored in the storage device 4, and the positional relationship between the reference points provided in the vehicle 1. From this, the current position of the vehicle 1 is calculated. The representative point indicating the position of the vehicle 1 is the origin of the vehicle coordinate system,
That is, the position of the vehicle, which is the center of the front wheel tires, is calculated as a value on a coordinate system (hereinafter referred to as a map coordinate system) having a fixed point determined at the point where the vehicle 1 starts moving as an origin. The calculated current position data of the vehicle 1 is input to the white line estimation device 7.

The white line estimating device 7 includes a past deviation data stored in the storage device 4 and a vehicle position calculating device 6.
Based on the current position of the vehicle 1 obtained in the above, the white line on the road is estimated as a function using the weight at the time of function estimation similarly stored in the storage device 4, and the deviation value is estimated based on this white line function. Do. Next, the actually measured deviation data obtained by the image processing device 3 is compared with the estimated deviation value, and the validity of the estimation of the deviation data is determined. Then, if the estimation is determined to be valid, the actually measured deviation data is stored in the storage device 3 and the actually measured deviation data is output, and if it is determined to be invalid, the estimated deviation value is output. .

Next, the operation of this embodiment will be described with reference to the flowcharts of FIGS. 4 to 6 and FIGS. (1) Photographing, White Line Data Extraction, Current Position Calculation In step S1, a road surface in front of the vehicle 1 is photographed by the imaging device 2. In step S2, based on the road surface photographing data obtained by the imaging device 2, the image processing device 3
By using a well-known image processing method, only the data of the white line extending in the traveling direction of the vehicle 1 is extracted. Further, since the white line image data is photographed in a state where the bird's eye is viewed from above the road surface, the image processing device 3 converts the white line image data into a planar state. The conversion method is well known, and its description is omitted.
Based on the forward observation point distance d _i stored white image extracted in step S3 the data and in the storage device 4, the image processing apparatus 3 is deviation y _i in these forward observation point distance d _i in the vehicle coordinate system values Is calculated as Or later,
The deviation data measured this time is represented by a subscript _i (i = 1 to 11: 1 is the data located closest to the vehicle).

In step S4, the current position of the vehicle 1 is obtained by the vehicle position calculating device 6 as a value on the map coordinate system. Then, the data indicating the deviation y _i and the current position of the vehicle 1 are input to the white line estimating device 7.

(2) Coordinate Conversion In step S5, the white line estimating device 7 converts the deviation data on the vehicle coordinate system into values on the map coordinate system. Specifically, as shown in FIG. 7, the current position of the vehicle 1 is set to (X ₀ ,
Y ₀ , Θ) and the coordinate value of the deviation on the vehicle coordinate system is (d _i , y _i ), the coordinate value (X _i , Y _i ) of the deviation on the map coordinate system is given by the following equation. . In the following description, variables expressed in uppercase letters are variables on the map coordinate system, and variables expressed in lowercase letters are variables on the vehicle coordinate system. X _i = X ₀ + d _i × cosΘ−y _i × sinΘ Y _i = Y ₀ + d _i × sinΘ + y _i × cosΘ If the coordinate transformation is performed in this way, the deviation data measured this time and stored in the storage device 4 Past deviation data can be handled on the same coordinate system.

(3) Data Deletion and Data Rearrangement In step S6, the forward fixation point distance di at which the deviation is measured is added to the deviation data subjected to coordinate conversion by the white line estimating device 7. The forward observation point distance d _i is used as a variable indicating the use limit of the data in subsequent steps. Thereafter, the deviation data is data (X _i , Y _i , d) obtained by adding the forward fixation distance to the coordinate value of the deviation on the map coordinate system.
_i ).

In step S7, the past deviation data stored in the storage device 4 is read by the white line estimating device 7, and the distance traveled by the vehicle 1 from the previous measurement to the present measurement is included in the deviation data. It is subtracted from the forward observation point distance d _i (i.e. utilization limits). In step S8, it is determined by the white line estimating device 7 whether or not this use limit value is smaller than a predetermined set limit value (5 [m] in this embodiment). The deviation data is deleted by the device 7, and if the determination is negative, the program proceeds to step S10.

[0022] In step S10, whether the data of the forward fixation point distance around d _i necessary for control by the white line estimation device 7 is determined. In the present embodiment, the deviation data for the forward fixation point distance d _i ± 0.5 [m] is calculated based on the forward fixation point distance d.
_Judge as data near _i .

In step S11, the past deviation data is rearranged by the white line estimating device 7. The sorting reference is the distance from the reference position in the vehicle 1 to the white line, and the sorting is performed sequentially from data near the vehicle 1 to data farther away. As a result, the convenience of use for calculating the estimation deviation function described later is achieved. Hereinafter, the past deviation data after the rearrangement is expressed by a subscript _j (j = 1,..., N: 1 is data located closest to the vehicle).

(4) Function Estimation In FIG. 4, when the program proceeds to step S12, a function estimation subroutine shown in the flowchart of FIG. 5 is started.

First, in step S31, the white line estimating device 7 selects four pieces of past deviation data closest to the vehicle 1. Next, in step S32, it is determined whether or not the deviation data is data in the vicinity of the forward fixation point distance. If the determination is affirmative, in step S33, the weight w _j at the time of function estimation described later is set to 1.2, and the determination is negative. If so, w _j = 0.8 is set in step S34.

Next, in step S35, the white line estimating device 7 calculates an estimation deviation function using the deviation data and the weight wj. In this embodiment, the deviation function is represented by f (X)
= Y = aX ² + bX + c is approximated as a quadratic function represented by a general formula. A, b when the sum of errors ε _j Σε _j = Σw _j (Y _j −aX _j ² −bX _j −c) ² (Σ is the sum from j = 1 to N) is minimized by the least squares method , C are constants to be obtained. ∂
From the simultaneous equations of Σε _j ² / ∂a, ∂Σε _j ² / ∂b, and ∂Σε _j ² / ∂c, aΣw _j ² X _j ⁴ + bΣw _j ² X _j ³ + cΣw _j ² X _j ² −Σw _j ² X _j ² Y _j = 0 aΣw _j ² X _j ³ + bΣw _j ² X _j ² + cΣw _j ² X _j −Σw _j ² X _j Y _j = 0 aΣw _j ² X _j ² + bΣw _j ² X _j + cNΣw _j = 0 I get it. By solving this simultaneous equation, the estimation deviation function f (X) = Y =
aX ² + bX + c is determined.

In step S36, the estimation deviation function f
The average value of the error between (X) = Y = aX ² + bX + c and the deviation data used to calculate the deviation function f (X) is calculated by the white line estimation device 7. In step S37, the average value of the errors is set to a predetermined setting error (in this embodiment, 5 [c
m]) It is determined whether it is within the range. If the determination is affirmative, the program proceeds to step S38, and if the determination is negative, the program proceeds to step S39.

In step S38, one piece of deviation data is added following the deviation data used for calculating the deviation function f. In this manner, as long as the average value of the errors is within the set error, the deviation data is sequentially added to calculate the deviation function f.

On the other hand, in step S39, the deviation function f is stored in the storage device 4 together with the range of the deviation data used for the calculation. Next, in step S40, among the deviation data used for calculating the deviation function f, four pieces of deviation data are newly selected from the data located farthest from the vehicle 1, and are newly selected for calculation of the deviation function f. Data. In this way, when the average value of the error is out of the range of the set error, the deviation data is newly selected and the deviation function f is calculated.

In step S41, it is determined whether or not the deviation function f has been calculated using all the past deviation data stored in the storage device 4. If the determination is affirmative, the program proceeds to the main routine of FIG. If the determination is negative, the process returns to step S32 to continue calculating the deviation function f. As described above, a plurality of deviation functions f within a predetermined set error range with respect to the deviation data can be calculated without overlapping the data.

(5) Deviation Estimation In FIG. 4, when the program proceeds to step S13, a deviation estimation subroutine shown in the flowchart of FIG. 6 is started. First, in step S51, the deviation data at the forward fixation point distance di is selected by the white line estimation device 7 from the deviation data measured this time. Then, in step S52, as shown in FIG. 8, line g along the measuring direction of this deviation data (x), and the straight line g (x) denote the extending direction of forward gaze point distance d _i line ( in other words, this intersection between a straight line along the traveling direction of the vehicle which passes through the reference point of the vehicle 1) beta is calculated based on the current position and forward observation point distance d _i of the vehicle 1. This intersection β indicates the position of the forward fixation point on the map coordinate system, as shown in FIG.

In step S53, a deviation function (in FIG. 8, function f ₂
Is selected by the white line estimating device 7. Hereinafter, as described later, a deviation function including past deviation data near the forward fixation point is referred to as a candidate function.

[0033] At step S54, the intersection point of the straight line g (x) and the candidate function f ₂ calculated in step S52 alpha is calculated by the white line estimation device 7, in step S55, the deviation data the intersection alpha is at function calculation (See step S39). If the determination is affirmative, the program proceeds to step S56, and if the determination is negative, the program proceeds to step S57. In the illustrated example, since the straight line g (x) has a candidate functions f ₂ and the intersection alpha, this intersection alpha present in the candidate function f ₂ in the range P ₁ to P ₂ of the deviation data at the time of calculation, the program steps S5
Move to 6. If the intersection α does not exist in the range P _{1 to} P ₂ , the program proceeds to step S57.

[0034] At step S56, the calculated distance between the point α and the point beta, that is, by the white line estimation unit 7 estimates of deviations in forward observation point distance d _i is based on the estimation deviation function (candidate functions) f ₂ Is done. Thereafter, the program proceeds to step S58.

In step S57, the candidate function is changed according to the position of the intersection point α. For example, in the example shown in FIG. 8, candidate functions if the intersection α is in front of P ₁ is f ₁
Is changed to, if located far from the P ₂ Conversely candidate function is changed to f _3. Thereafter, the program returns to step S54.

In step S58, the estimated value of the deviation at the forward fixation point distance d calculated in step S56 is compared with the deviation data at the forward fixation point distance d selected in step S51. It is determined whether or not the difference from the estimated value is within a predetermined error (referred to as a control limit; in this embodiment, 10 cm). If the determination is affirmative, the deviation data is output from the white line estimating device 7 in step S59, and if the determination is negative, the estimated value of the deviation is output in step S60.

(6) Data deletion In step S14, the deviation data determined not to be within the control limit in step S58 is deleted, and the remaining deviation data is stored in the storage device 4 to estimate the next deviation function. Used for Thereafter, the program returns to step S1, and the deviation is measured again. Then, by performing the above procedure at every predetermined control cycle, traveling data used for steering of the vehicle or the like is periodically output.

Therefore, according to the present embodiment, of the measured deviation data, all data useful for calculating the estimation deviation function f are stored in the storage device 4, and the estimation deviation function is calculated using these deviation data. Since the calculation of f is performed, even if a defect occurs in the deviation data at the time of measurement, the data can be accurately estimated by using useful and large number of deviation data stored in the storage device 4. In addition, since data near the forward fixation point distance d _i is weighted when the estimation deviation function is calculated, data estimation near the forward fixation point distance d, which is the most important for controlling the autonomous vehicle 1, is improved. It can be performed with high accuracy.

The details of the traveling data computing device for an autonomous traveling vehicle according to the present invention are not limited to the above-described embodiment, and various modifications are possible.

[0040]

As described above in detail, according to the present invention, only when the calculated coordinate value of the reference line is within a predetermined error range with respect to the reference line function, this coordinate value is stored in the storage means. Since they are stored, only the coordinate values useful for calculating the reference line function are all stored in the storage means. Therefore, when calculating the reference line function using the coordinate values in the storage means, even if the coordinate value of the reference line is lost at the time of measurement, the coordinates are calculated based on the useful and numerous coordinate values stored in the storage device. The value can be accurately estimated.

[Brief description of the drawings]

FIG. 1 is a diagram corresponding to claims of the present invention.

FIG. 2 is a block diagram showing an autonomous traveling vehicle to which the traveling data calculation device according to one embodiment of the present invention is applied.

FIG. 3 is a diagram showing a relationship between a vehicle coordinate system and a deviation.

FIG. 4 is a flowchart for explaining the operation of one embodiment.

FIG. 5 is a flowchart illustrating a function estimation subroutine.

FIG. 6 is a flowchart illustrating a deviation estimation subroutine.

FIG. 7 is a diagram for explaining coordinate conversion from a vehicle coordinate system to a map coordinate system.

FIG. 8 is a diagram for explaining a procedure for calculating an estimated value of deviation data.

[Explanation of symbols]

 a imaging means b image processing means c coordinate calculation means d storage means e function calculation means f determination means 1 vehicle 2 imaging device 3 image processing device 4 storage device 5 rotary encoder 6 vehicle position calculation device 7 white line estimation device

Claims (1)

(57) [Claims] 1. An image pickup means for photographing a road surface in front of a vehicle and outputting image data at predetermined intervals such that an overlapped portion is formed in an adjacent image; Image processing means for extracting an extended reference line; coordinate calculation means for calculating the coordinate value from the reference line extracted by the image processing means; storage in which coordinate values of the reference line calculated in the past are stored Means, a function calculating means for calculating a reference line function that approximates the coordinate values based on the coordinate values in the storage means, a coordinate value obtained by the coordinate calculating means, and a reference obtained by the function calculating means. An error from the line function is determined, and if the error is within a predetermined range, the coordinate value is output as traveling data and the coordinate value is stored in the storage means. line Traveling data arithmetic unit autonomous vehicle, characterized in that a determination means for outputting the coordinate values of the reference line is estimated from the number as the travel data.