JP5867340B2 - Road curvature calculator - Google Patents

Description

  The present invention relates to a travel path curvature calculation apparatus that extracts a lane line that divides a travel path of the vehicle from a road surface image in front of the vehicle and calculates a curvature of the travel path based on the lane line.

  2. Description of the Related Art Conventionally, a so-called white line recognition technique has been proposed in which a road surface image in front of a vehicle is picked up by a camera or the like, and a lane line that divides the traveling path of the vehicle, such as a white line, is extracted from the road surface image. Also, when the vehicle is traveling on a curved road, a white line is similarly extracted so that when the lateral deviation of the vehicle with respect to the center of the traveling path exceeds a predetermined value, the vehicle is returned to the center of the traveling path. It has also been proposed to generate a yaw moment by braking (see, for example, Patent Document 1).

JP 2000-33860 A

  Here, when the vehicle is traveling on a curved road, the curvature of the traveling road may be calculated based on the lane markings and applied to the control as described above. It is desirable to apply filtering to the curvature. The filter processing here is processing for correcting the calculated curvature and suppressing instantaneous fluctuation of the curvature. By doing this, even if the calculated value of the curvature fluctuates instantaneously, for example, by misdetecting the lane marking, the correction is suppressed by the correction, and the control such as braking is performed. Unstable execution can be suppressed.

  However, in such a filtering process, if a filtering process that greatly suppresses the variation in the calculated curvature is executed, the control using the curvature is reduced in accuracy or delayed. On the other hand, when the lane marking is a discontinuous line, the calculated curvature is more likely to vary than when it is a continuous line, and it is necessary to significantly suppress the calculated variation in curvature.

  Therefore, the present invention provides a traveling road curvature calculation apparatus capable of suppressing instantaneous fluctuations in curvature by correcting the curvature of the traveling road calculated based on the lane lines extracted from the road surface image. It was made for the purpose of appropriately performing the correction depending on whether the line is a continuous line or a discontinuous line.

  In the traveling road curvature calculation apparatus of the present invention made to achieve the above object, the lane line extraction means extracts a lane line that divides the traveling road of the vehicle from the road surface image in front of the vehicle acquired by the road surface image acquisition means. To do. Then, the line type determination means determines whether the lane line extracted by the lane line extraction means is a continuous line or a discontinuous line, and the curvature calculation means determines the lane line extracted by the lane line extraction means. Based on this, the curvature of the travel path is calculated.

  The curvature correction means corrects the curvature calculated by the curvature calculation means so that instantaneous fluctuation of the curvature is suppressed. Here, the correction mode of the curvature correction unit is changed as follows by the correction mode change unit. That is, the variation is greatly suppressed when the line type determining unit determines that the line type determining unit is a discontinuous line, compared to when the line type determining unit determines that the line type is a continuous line. Be changed.

  For this reason, when the lane line is a discontinuous line and the calculated curvature is likely to vary compared to the case where the lane line is a continuous line and the calculated curvature does not vary so much, the curvature correction means The variation is greatly suppressed. Therefore, in the present invention, the correction can be appropriately performed according to whether the dividing line is a continuous line or a discontinuous line.

It is a block diagram showing the structure of the traveling path curvature calculation apparatus with which this invention was applied. It is a flowchart showing the process of the control part of the travel path curvature calculation apparatus. It is explanatory drawing showing the structure of the map used by the process. It is explanatory drawing explaining the parameter of the horizontal axis of the map. It is explanatory drawing which illustrates the effect of the said process with respect to the solid white line. It is explanatory drawing which illustrates the effect of the said process with respect to the broken white line.

[Configuration of the embodiment]
Next, an embodiment to which the present invention is applied will be described with reference to the drawings. As shown in FIG. 1, a traveling path curvature calculation apparatus 1 as an embodiment of the present invention includes an in-vehicle camera 3 and a control unit 5. The in-vehicle camera 3 is attached in the vicinity of the base of the rear side of the rear-view mirror at the center upper portion of the windshield of the vehicle (not shown) on which the traveling road curvature calculation device 1 is mounted, and takes a road surface image in front of the vehicle. The image data is input to the control unit 5. The control unit 5 is configured as a microcomputer including a CPU, a ROM, and a RAM, recognizes a white line (an example of a division line) on the road based on image data input from the in-vehicle camera 3, and the vehicle is operated as follows. Calculate the curvature of the driving lane (an example of a driving path).

[Processing in Embodiment]
The process shown in the flowchart of FIG. 2 is repeatedly executed by the CPU of the control unit 5 at predetermined timings (for example, every 16 ms) based on the program stored in the ROM of the control unit 5. As shown in FIG. 2, in this process, first, in S1 (S represents a step: the same applies hereinafter), white lines on both the left and right sides of the lane in which the vehicle is traveling are obtained from the road surface image taken by the in-vehicle camera 3. The extracted white line recognition process is executed, and the curvature of the lane is acquired (calculated) from the extracted white line. A method of acquiring the curvature of the lane from the extracted white line position is well known and will not be described in detail here. In the following description, the curvature is represented by the reciprocal of the radius of curvature (unit: m ⁻¹ ).

  In subsequent S2, it is determined whether or not the white lines on both the left and right sides extracted in S1 are solid lines. A method for determining whether a white line extracted from a road surface image is a solid line (an example of a continuous line) or a broken line (an example of a discontinuous line) is disclosed in, for example, Japanese Patent Application Laid-Open No. 2003-178399. A known method can be applied. If the white lines on both the left and right sides are both solid lines (S2: Y), the process proceeds to S3, where both-side solid line filter coefficient maps are set.

  Here, in this embodiment, a limit rate (an example of an allowable value) is used as a filter coefficient, and the limit rate used when the white lines on both the left and right sides are both solid lines is shown in FIG. It is determined by the map (Both-side solid line limit rate determination map). Note that the vertical axis in FIG. 3A is the limit rate, which was acquired in S1 this time with respect to the value obtained by correcting the curvature acquired in the previous process using this map (the value after the previous limit). It is instructed to correct the curvature so as to be within the range of fluctuation defined by the limit rate. Also, the horizontal axis in FIG. 3 is the absolute value of the average difference between two successive sections, and is calculated by the following equation (1).

However, in a few 1, X is a curvature obtained in S1, rate songs X ₀ is acquired in this S1, X over _i is the curvature obtained in S1 of i times before the treatment (limit rate by uncorrected ). M and N represent the time window width for obtaining the average in terms of the number of samples, and correspond to the time window width (msec) divided by the sample time (that is, the processing cycle of the process in FIG. 2). Furthermore, LagSampNum represents the deviation width of two time windows for obtaining an average by the number of samples. Therefore, when M = N = LagSampNum, the arrangement of M, N, and LagSampNum with respect to the current time t can be imagined as illustrated in FIG. Assuming that M, N, and LagSampNum have widths corresponding to 1000 ms, respectively, when the processing cycle is 16 msec, 1000 / 16≈63 (samples). Further, in order to increase the responsiveness, it is effective to reduce LagSampNum as illustrated in FIG. 4B, and when LagSampNum corresponds to 500 ms, 500 / 16≈31 (sample).

  Note that the values of M, N, and LagSampNum are preset and stored in the ROM of the control unit 5. Further, the value of Equation 1 is a small value for a steady curve or straight line, and is also an index indicating how much the curvature of the lane actually fluctuates.

  Returning to FIG. 2, when the double-side solid line filter coefficient map of FIG. 3A is set in S3 in this way, in S4, the limit rate as the filter coefficient is determined using the map. Is done. That is, based on the curvature acquired in S1 of the current process and the curvature acquired in S1 of the previous process and stored in the RAM of the control unit 5, the value of Equation 1 is calculated. The limit rate corresponding to the value is determined based on the map of FIG. In subsequent S5, the curvature (camera output curvature value) acquired in S1 of the current process is filtered. In other words, the curvature acquired in S1 of the current process is corrected so that the variation of the curvature acquired in S1 of the previous process with respect to the corrected value (post-limit value) is within the limit rate range even at the maximum. The In S6, the curvature is transmitted to another control unit or the like, so that it can be used for a process such as a known lane departure prevention system, and the process is temporarily ended.

  On the other hand, when the white lines on both sides are not solid lines (S2: N), it is determined in S7 whether only one side of the white lines on the left and right sides is a broken line. If only one side is a broken line (S7: Y), the process proceeds to S8, and the one-side broken line filter coefficient map (one-side broken line limit rate determination map) shown in FIG. 3B is set and processed. Shifts to S4 described above. If only one side is not a broken line (S7: N), that is, if the white lines on the left and right sides are broken lines, the process proceeds to S9, and the filter coefficient map for both-side broken lines shown in FIG. The limit rate determination map) is set, and the process proceeds to S4 described above.

[Effect of the embodiment]
As shown in FIGS. 3A, 3 </ b> B, and 3 </ b> C, in any map, the limit rate corresponding to the map increases as the value of Equation 1 (the absolute value of the average difference between the two sections) increases. Further, the following points are common. That is, each map is configured such that the limit rate becomes 0.005 ( m ⁻¹ ) when the value of Equation 1 becomes 0.01 ( m ⁻¹ ). Also, until the limit rate increases to 0.0005 ( m ^-1 ), the increase of the limit rate with respect to the increase in the value of Equation ¹ is gradual, and from the vicinity where the limit rate becomes 0.0005 ( m ^-1 ), The rate of increase of the limit rate with respect to the increase of the value of Equation 1 is increasing rapidly. However, the position of the inflection point where this sudden increase starts (in other words, the position where the value obtained by differentiating the limit rate twice with the value of Formula ¹ is the maximum) is 0.0008 ( m ^-1 ) in (A), (B) is different from 0.0015 (m ⁻¹ ) and (C) is 0.002 ( m ⁻¹ ).

For this reason, when the white lines on both the left and right sides are solid lines (S2: Y), by using the map of FIG. 3A, the value of Equation ¹ exceeds 0.0008 ( m ^-1 ), The change in curvature (hereinafter also referred to as a calculated curvature value) acquired in S1 is greatly reflected in the curvature (post-limit value) used for control in S6. On the other hand, when the white line on one side is a broken line (S7: Y), since the value of Equation ¹ exceeds 0.0015 ( m ^-1 ), the change in the calculated curvature value is greatly reflected in the post-limit value. If the white lines on both the left and right sides are broken lines (S7: N), the change in the calculated curvature value is greatly reflected in the value after the limit when the value of Equation ¹ exceeds 0.002 ( m ^-1 ). Will come to be. That is, compared to the case where the left and right white lines are solid lines (S2: Y), when the white line on one side is a broken line (S7: Y), the fluctuation of the post-limit value is greatly suppressed, and the white lines on both left and right sides Is a broken line (S7: N), the fluctuation of the post-limit value is further greatly suppressed.

  Here, the calculated curvature value obtained in S1 indicates that false detection or the like occurs when the white line on one side is a broken line (S7: Y) compared to when the white line on both the left and right sides is a solid line (S2: Y). If the white line on both the left and right sides is a broken line (S7: N), it is more likely to fluctuate. If the curvature that has fluctuated due to erroneous detection or the like is used as it is in the processing of the lane departure prevention system, for example, when the steering angle is controlled to prevent lane departure, the steering is controlled to vibrate. The driver may feel uncomfortable. Therefore, as described above, the filter processing that suppresses the fluctuation of the post-limit value is effective. However, if the filter is made too large to suppress the fluctuation of the post-limit value, the control is reduced in accuracy or delayed. .

  Therefore, in this embodiment, depending on whether each white line on the left and right is a solid line or a broken line, the greater the curvature calculation value is likely to fluctuate due to the influence of false detection, etc. It is greatly suppressed. Therefore, in the present embodiment, the calculated curvature value can be appropriately corrected depending on whether the left and right white lines are solid lines or broken lines.

FIG. 5 is a diagram illustrating the effect of the processing on a lane in which the left and right white lines are solid lines, and FIG. 6 is a diagram illustrating the effect of the processing on a lane in which the left and right white lines are broken lines. In each figure, the horizontal axis represents time (one scale is 10 seconds), and the vertical axis represents curvature (one scale is 1 × 10 ⁻³ m ⁻¹ ). When the white lines on both the left and right sides are solid lines (S2: Y), there is little influence of false detection and the like, and the variation in the calculated curvature value is small as shown by the thick line in FIG. For this reason, even if the fluctuation of the calculated curvature value is not so much suppressed (S3), a post-limit value that favorably reflects the curvature of the lane can be obtained as shown by a thin line in FIG.

  On the other hand, when the white lines on both the left and right sides are broken lines (S7: N), the influence of false detection or the like is large, and the calculated curvature value varies greatly as shown by the thick line in FIG. Therefore, in the present embodiment, by setting a filter coefficient map for both-side broken lines to greatly suppress the fluctuation of the curvature calculation value (S9), as shown by a thin line in FIG. 6, although there is a slight delay, A post-limit value that reflects the curvature well is calculated. As described above, according to the present embodiment, it is possible to appropriately correct the curvature calculation value according to whether each of the left and right white lines is a solid line or a broken line, and to control the white line departure prevention system and the like satisfactorily. . The white line departure prevention system may execute a Lane Departure Warning (LDW) control that only generates an alarm, and prevents white line departure by steering or braking (LDP). : Lane Departure Prevention) Control may be executed.

[Other Embodiments]
In the above embodiment, the in-vehicle camera 3 corresponds to a road surface image acquisition unit, and the control unit 5 includes a lane marking extraction unit, a line type determination unit, a curvature calculation unit, a curvature correction unit, a correction mode change unit, and an allowable value calculation unit. , And fluctuation range adjusting means. Of the processing of the control unit 5, S1 is a lane marking extraction means and curvature calculation means, S2 and S7 are line type determination means, S4 and S5 are curvature correction means, and S3, S8, and S9 are correction mode changes. S4 corresponds to the allowable value calculation means, and S5 corresponds to the fluctuation range adjustment means.

  The present invention is not limited to the above-described embodiment, and can be implemented in various forms without departing from the gist of the present invention. For example, various known filter processes can be applied as a process for correcting the calculated value of curvature. It may be a simple filter that simply takes the average value of the most recent N data. In that case, the value of N may be changed depending on whether the white line is a solid line or a broken line. For example, if one side is a broken line, N may be changed to N + α (0 <α), and if both sides are broken lines, N may be changed to N + β (α <β). Further, the discontinuous line may be further divided into blocks and broken lines, and the filtering process may be further subdivided, or the filtering process may be further divided for double lines and yellow lines. On the other hand, when only one side is a broken line, the same filtering process may be executed as when both sides are solid lines or both sides are broken lines.

  Furthermore, in each of the above embodiments, the filter processing is performed on the curvature of the travel road calculated from the white lines on both the left and right sides, but the curvature of the white lines on both the left and right sides is considered as the curvature of the travel road for convenience. The filter process may be executed at, and the curvature of the traveling road may be calculated from the respective curvatures after the filter process. Further, the curvature of the traveling road may be calculated with reference to only one of the left and right white lines. The configurations of the above embodiments can be arbitrarily combined.

[Various forms of the present invention]
However, as described above, the curvature calculation unit calculates the curvature at every predetermined timing, and the curvature calculation unit calculates N consecutive times (N is an integer of 2 or more) including the latest timing. The average value of the calculated curvature is the latest average value, and the average value of the curvature calculated at the timing of M times (M is an integer of 2 or more) at least before the latest timing is the past average value. The curvature correcting means includes an allowable value calculating means for calculating an allowable value of the fluctuation range of the curvature based on an absolute value of a difference between the latest average value and the past average value, and the curvature calculating means The curvature calculated at the latest timing is the current value, calculated by the curvature calculation means at the timing immediately before the latest timing, and corrected by the curvature correction means. A fluctuation range in which the fluctuation range from the previous value after the correction of the current value is included in the allowable value calculated by the allowable value calculation means by correcting the current value as the immediately previous value after correction of the curvature after Adjusting means, and the correction mode changing means has the following actions and effects when changing the correspondence between the allowable value calculated by the allowable value calculating means and the absolute value. That is, in this case, the correction mode of the correction unit can be changed in more detail and appropriately.

  Further, in this case, the correspondence relationship between the allowable value and the absolute value is such that the allowable value corresponding to the absolute value increases as the absolute value increases, and the increase ratio of the allowable value with respect to the increase of the absolute value. May include an inflection point that rapidly increases, and the correction mode changing means may change the position of the inflection point. In this case, the correspondence relationship between the allowable value and the absolute value can be changed more appropriately.

  In addition, when the lane marking extraction unit extracts the lane markings on both sides of the travel path, the line type determination unit determines whether the lane markings on the both sides are the continuous lines or the discontinuities. The correction mode changing means determines whether one of the lane markings is compared with the case where the line type determining means determines that each lane marking on both sides is the continuous line. In the case where the line type determination means determines that the other division line of the continuous line is the discontinuous line, the variation correction by the curvature correction means is greatly reduced, and either Compared to the case where the line type determining means determines that the lane marking is the continuous line and the other lane marking is the discontinuous line, the line is determined that each lane marking on both sides is the discontinuous line. The curvature correction is more effective when the seed judging means judges. As the suppression of the fluctuation due to stage is made substantially, it may change the correction mode of the curvature correction device. In this case, the correction mode can be changed in more detail and appropriately depending on the degree to which the calculated curvature is likely to vary.

DESCRIPTION OF SYMBOLS 1 ... Traveling path curvature calculation apparatus 3 ... Car-mounted camera 5 ... Control part

Claims (4)

Road surface image acquisition means (3) for acquiring a road surface image in front of the vehicle;
Lane line extraction means (5, S1) for extracting a lane line that divides the travel path of the vehicle from the road surface image acquired by the road surface image acquisition means;
Line type determination means (5, S2, S7) for determining whether the lane line extracted by the lane line extraction means is a continuous line or a discontinuous line;
Curvature calculation means (5, S1) for calculating the curvature of the travel path based on the lane line extracted by the lane line extraction means;
Curvature correction means (5, S4, S5) for correcting the curvature calculated by the curvature calculation means and suppressing instantaneous fluctuations in the curvature;
Compared to the case where the line type determining unit determines that the line is a continuous line, the variation correction by the curvature correcting unit is significantly performed when the line type determining unit determines that the line is a discontinuous line. Correction mode changing means (5, S3, S8, S9) for changing the correction mode of the curvature correction means,
A travel path curvature calculating apparatus comprising: The curvature calculation means calculates the curvature every predetermined timing,
An average value of the curvatures calculated at N consecutive times (N is an integer of 2 or more) including the latest timing by the curvature calculation means is set as the latest average value, and at least before the latest timing. The average value of the curvatures calculated at successive M times (M is an integer of 2 or more) is a past average value.
The curvature correction means includes
An allowable value calculation means (5, S4) for calculating an allowable value of the variation range of the curvature based on an absolute value of a difference between the latest average value and the past average value;
The curvature calculated at the most recent timing by the curvature calculating means is a current value, calculated by the curvature calculating means at a timing immediately before the most recent timing, and corrected by the curvature correcting means. Fluctuation width adjustment means for correcting the current value as a value immediately before correction after correction, so that a fluctuation width of the current value from the previous value after correction falls within the allowable value calculated by the allowable value calculation means ( 5, S5),
With
The travel path curvature calculation apparatus according to claim 1, wherein the correction mode changing unit changes a correspondence relationship between the allowable value calculated by the allowable value calculating unit and the absolute value. The correspondence relationship between the tolerance value and the absolute value indicates that the tolerance value corresponding to the absolute value increases, and the tolerance value corresponding thereto increases, and the value obtained by differentiating the tolerance value twice with the absolute value in the correspondence relationship is as follows. With the largest inflection point,
The travel path curvature calculation apparatus according to claim 2, wherein the correction mode changing unit changes the position of the inflection point. When the lane marking extraction unit extracts the lane markings on both sides of the travel path, the line type determination unit determines whether the lane markings on the both sides are the continuous lines or the discontinuous lines. Judgment is made for each,
Compared to the case where the line type determining unit determines that each lane marking on both sides is the continuous line, the correction mode changing unit is configured such that one of the lane markings is the continuous line and the other lane marking. When the line type determining means determines that is a discontinuous line, the curvature correction means largely suppresses the fluctuation, and one of the dividing lines is the continuous line and the other is the other. Compared with the case where the line type determination means determines that the lane marking is the discontinuous line, the case where the line type determination means determines that the lane markings on both sides are the discontinuous lines. The traveling path curvature according to any one of claims 1 to 3, wherein a correction mode of the curvature correction unit is changed so that the variation is largely suppressed by the curvature correction unit. Computing device.