JP2007011492A - Driving lane detection apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a driving lane detection apparatus capable of highly accurately finding out the curvature of a driving lane. <P>SOLUTION: In the driving lane detection apparatus for photographing a driving lane by an image pickup means and finding out the curvature of the driving lane, the curvature of the driving lane on a preceding position from a present position is found out, and when the curvature of the driving lane on the preceding position is larger than that of the present position, a change in the curvature of the driving lane is permitted as compared with a smaller case, and the curvature of the driving lane on the present position is found out on the basis of the permitted curvature change. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a travel path detection device that captures an image of a travel path by an imaging unit and obtains a curvature of the travel path.

The lane keeping device captures the front of the vehicle with a camera, recognizes the lane from the captured image of the travel path, and adds steering torque to the steering mechanism so that the vehicle travels along the center of the lane (see Patent Document 1). . The lane keeping device obtains the curvature of the road (curve radius) from the relationship between the recognized lane and the vehicle, the yaw angle of the vehicle with respect to the lane, the offset of the vehicle from the lane center, etc., and additional steering torque based on these parameters Set. Therefore, in order to improve the accuracy with which the vehicle travels along the center of the lane, it is important to obtain parameters such as the curvature of the travel path with high accuracy.
JP 2001-10518 A

  However, the curvature calculated from the captured image includes noise, and the curvature fluctuates violently in an instant. In particular, the calculated curvature fluctuates even when the actual road curvature does not change during straight running. As a result, the accuracy of lane keeping also decreases. As a cause of noise included in this curvature, the accuracy of the camera and the recognition accuracy of the lane (a pair of white lines) in the image processing for the captured image can be considered. In image processing, white lines provided at regular intervals are detected, and the detected white lines are connected to obtain the curvature. If there is noise on the captured image, or if the white line is thin or missing, the white line cannot be detected accurately. For this reason, the detected white line cannot be connected along the actual white line. As a result, an accurate curvature cannot be obtained, and the obtained curvature varies.

  Then, this invention makes it a subject to provide the traveling path detection apparatus which can obtain | require the curvature of a traveling path with high precision.

  The travel path detection device according to the present invention images a travel path by an imaging unit and determines the curvature of the travel path in a travel path detection apparatus that calculates the curvature of the travel path. When the curvature of the road at the position is large, the change in the curvature of the road is allowed compared to when the curvature is small, and the curvature of the road at the current position is obtained based on the allowed change in curvature. .

  In this travel path detection device, the travel path is imaged, and the curvature of the travel path is obtained from the captured image. At this time, the travel path detection device sequentially obtains the curvature of the travel path at a position before the current position in order to acquire the shape of the road that has traveled in the past. In the travel path detection device, the curvature of the travel path is allowed to change as the curvature of the travel path at a position before the current is larger. In other words, the sharper the curve, the larger the change in the curve radius, so that a large change in curvature is allowed. On the other hand, the closer to a straight line or the closer the straight line, the smaller the change in curve radius (since there is no change in the case of a straight line), so only a small change in curvature is allowed (the change in curvature is limited). Further, the travel path detection device obtains the curvature of the travel path at the current position within the allowable range of curvature change. In this way, by changing the amount of change from the curvature obtained previously according to the road shape, the change in curvature is suppressed when traveling on a straight line, and the large curvature change is performed when traveling on a curve. Can be tolerated. Therefore, noise included in the curvature detected from the captured image can be suppressed (particularly in the case of a straight line), and the curvature that follows the actual curvature can be obtained in the curve. As a result, with this travel path detection device, the curvature of the travel path can be obtained with high accuracy.

  According to the present invention, the curvature of the traveling road can be obtained with high accuracy.

  Hereinafter, an embodiment of a travel path detection device according to the present invention will be described with reference to the drawings.

  In the present embodiment, the traveling path detection device according to the present invention is applied to a lane keeping device. The lane keeping device according to the present invention recognizes a white line from an image captured by a camera and assists steering by a driver, and applies an auxiliary steering torque toward the center of left and right white lines (lanes).

  A lane keeping device 1 according to the present embodiment will be described with reference to FIGS. FIG. 1 is a configuration diagram of a lane keeping device according to the present embodiment. FIG. 2 is a control block diagram of the lane keeping device according to the present embodiment. FIG. 3 is a diagram showing the relationship between the travel distance and the curvature of the travel path.

  The lane keeping device 1 sets a target steering torque necessary for assisting traveling in the center of the lane, and applies the target steering torque by the electric power steering device. At that time, the lane keeping device 1 sets the target steering torque based on the curvature of the traveling path, the vehicle direction (yaw angle) with respect to the white line, and the deviation (offset) of the vehicle position with respect to the lane center. In particular, the lane keeping device 1 sets a maximum change amount (gradient limit) from the previously obtained curvature according to the road shape (curvature) that has traveled in the past, and obtains a corrected curvature based on the maximum change amount. . The lane keeping device 1 includes a white line recognition sensor 2 and a lane keeping ECU [Electronic Control Unit] 3, and uses an electric power steering device 10. The white line recognition sensor 2 includes a camera 2a as an imaging unit and a white line recognition ECU 2b.

  The electric power steering device 10 controls the EPS motor 10b by an EPS [Electronic Power Steering] ECU 10a, and assists the steering by the driver with the driving torque of the EPS motor 10b. As shown in FIG. 2, the EPS ECU 10a sets an assist torque based on the steering torque generated by the driver's steering, and drives and controls the EPS motor 10b according to the assist torque. In particular, when the EPSECU 10a receives the steering torque signal TS from the lane keep ECU 3, the EPS ECU 10a multiplies the target steering torque indicated by the steering torque signal TS by a predetermined coefficient, and adds the multiplied value (additional torque) to the assist torque. The EPS motor 10b is driven and controlled according to the assist torque + addition torque. Driving torque by the EPS motor 10b is added to the steering mechanism, and torque by the EPS motor 10b is applied to the steering mechanism in addition to the steering torque by the driver. The additional torque is a relatively small torque that assists the steering torque by the driver.

  The camera 2a is, for example, a CCD [Charge Coupled Device] camera, and is attached to the front of the vehicle on which the lane keeping device 1 is mounted. At this time, the camera 2a is attached so that the optical axis direction thereof coincides with the traveling direction of the vehicle. The camera 2a captures a road ahead of the vehicle and acquires a captured color image (for example, an image of RGB [RedGreen Blue]). The camera 2a transmits the captured image data to the white line recognition ECU 2b as an imaging signal PS. The camera 2a has a wide imaging range in the left-right direction, and can sufficiently capture the white lines on both the left and right sides (a pair) indicating the traveling lane. Although the camera 2a is colored, it may be a black and white camera as long as it can acquire an image that can recognize a white line on the road.

  The white line recognition ECU 2b includes a CPU [Central Processing Unit], a ROM [Read Only Memory], a RAM [Random Access Memory], and the like. The white line recognition ECU 2b takes in the imaging signal PS and recognizes a pair of white lines indicating the lane in which the vehicle is traveling from the captured image data of the imaging signal PS. Then, the white line recognition ECU 2b calculates a lane width and a line passing through the center of the pair of white lines (that is, the center of the lane) from the recognized pair of white lines. Further, the white line recognition ECU 2b calculates the radius of the lane center (curve radius R), and calculates the curvature γ (= 1 / R) from the curve radius R. Further, the white line recognition ECU 2b calculates the vehicle direction (yaw angle θ) with respect to the white line and the position of the vehicle center (offset D) with respect to the center of the lane. Then, the white line recognition ECU 2b transmits the information of the recognized pair of white lines and the calculated information to the lane keep ECU 3 as an image signal GS.

  The lane keep ECU 3 includes a CPU, a ROM, a RAM, and the like. The lane keep ECU 3 takes in the image signal GS from the white line recognition sensor 2 at regular intervals. In the lane keeping ECU 3, when the lane keeping device 1 is activated by an operation by the driver, various information (curvature γ, yaw angle θ, and the like shown in the image signal GS is set so that the vehicle travels near the center of the lane. A target steering torque is set based on the offset D), and a steering torque signal TS indicating the target steering torque is transmitted to the electric power steering apparatus 10 (EPS ECU 10a). The lane keep ECU 3 executes various processes (curvature correction process, target steering torque setting process, etc.) by executing an application program (software) for the lane keep process.

  As shown in FIG. 2, the lane keep ECU 3 performs a curvature correction process on the curvature γ detected by the white line recognition sensor 2 at regular intervals to obtain a corrected curvature γ ′. Then, the lane keep ECU 3 multiplies the corrected curvature γ ′ by the gain G1, multiplies the yaw angle θ by the gain G2, and multiplies the offset D by the gain G3 by a target steering torque setting process at regular intervals. A target steering torque is set based on the corrected curvature γ ′ multiplied by each gain G1 to G3, the yaw angle θ, and the offset D. Here, for the corrected curvature γ ′, the target steering torque is set such that the vehicle follows the road curve, the yaw angle θ converges to 0, and the offset D converges to 0. Is done. The curvature correction process will be described in detail below.

  In the lane keep ECU 3, the detected curvature γ is greater than the curve determination threshold γ0, the curvature γ is equal to or less than the curve determination threshold γ0 and greater than the curve determination threshold γ1, or the curvature γ is equal to or less than the curve determination threshold γ1 at regular time intervals. Determine whether. The curve determination threshold values γ0 and γ1 are threshold values for determining the degree of curve of the traveling road, and have a relationship of γ0> γ1. When the curve determination threshold value γ0 is larger, the curve has a relatively small curve radius. When the curve determination threshold value γ0 is smaller than or equal to the curve determination threshold value γ1, the curve has a relatively larger curve radius. In some cases it is straight or nearly straight.

  In the lane keep ECU 3, when it is determined that the curvature γ is larger than the curve determination threshold γ0, when the curvature γ is determined to be smaller than the curve determination threshold γ0 and larger than the curve determination threshold γ1, or the curvature γ is equal to or smaller than the curve determination threshold γ1. When the determination is made, the number of times each determination result continues is counted. Then, the lane keep ECU 3 determines whether or not the count number exceeds the timer threshold value τ. The timer threshold value τ is a threshold value for determining whether or not a sufficient time has passed to determine the road shape that has traveled in the past. Here, it is determined whether a relatively small curve radius curve, a relatively large curve radius curve, or a straight line or almost a straight line has been continuously traveled from the present time for the past predetermined time (for example, about several seconds). . As a result, the shape of the road that has traveled in the past (straight or curved, or the degree of curve in the case of a curve) is determined. Even if each determination condition is met instantaneously, the curvature γ detected by the white line recognition sensor 2 may contain noise, so the road shape is not fixed unless it continues for a predetermined time.

  When the number of times that the curvature γ is continuously determined to be greater than the curve determination threshold γ0 exceeds the timer threshold τ, the lane keep ECU 3 sets α0 as the gradient limit α. When the number of times of continuous determination that the curvature γ is equal to or less than the curve determination threshold γ0 and greater than the curve determination threshold γ1 exceeds the timer threshold τ, the lane keep ECU 3 sets α1 as the gradient limit α. When the number of times that the curvature γ is continuously determined to be equal to or less than the curve determination threshold γ1 exceeds the timer threshold τ, the lane keep ECU 3 sets α2 as the gradient limit α. The gradient limit α is the maximum allowable change in curvature from the previous corrected curvature γ ″ to the current corrected curvature γ ′, and the initial value is α0. Α0 is a curve having a relatively small curve radius. The maximum value is set so as to follow the change in curvature on a steep curve, and at least a value smaller than the value obtained from the road design standard and the vehicle speed is set. Is a slope limit value corresponding to a curve having a relatively large curve radius, and is set to a value smaller than α0 and greater than α2, where α2 is a slope limit value corresponding to a straight line or almost a straight line. Is set to the smallest value so that the change in curvature can be suppressed.

  The lane keep ECU 3 then determines whether or not the absolute value of the value obtained by subtracting the previous corrected curvature γ ″ from the detected curvature γ is greater than the gradient limit α (that is, the curvature γ from the previous corrected curvature γ ″). It is determined whether or not the amount of change is greater than the slope limit α. When the absolute value of the value obtained by subtracting the previous corrected curvature γ ″ from the curvature γ is larger than the gradient limit α, the curvature γ has changed from the previous corrected curvature γ ″ to be larger than the gradient limit α. Therefore, in order to suppress the change within the gradient limit α, the lane keep ECU 3 adds the gradient limit α to the previous corrected curvature γ ″ when the curvature increases to obtain a corrected curvature γ ′. Is decreased, the gradient limit α is subtracted from the previous corrected curvature γ ″ to obtain a corrected curvature γ ′. On the other hand, when the absolute value of the value obtained by subtracting the previous corrected curvature γ ″ from the curvature γ is equal to or less than the gradient limit α, the curvature γ changes within the gradient limit α from the previous corrected curvature γ ″. Therefore, the lane keep ECU 3 sets the curvature γ detected this time as it is as the corrected curvature γ ′.

  In this way, the road shape that has traveled for a predetermined time until reaching the current position is estimated from the curvature detected every moment, and the maximum allowable change amount of the curvature used in the control is set in three stages according to the road shape. . As a result, the curvature used for the control in the curve can change following the actual curvature of the curve, and noise is removed from the curvature used for the control in the straight line.

  FIG. 3 is an example of a change in curvature when traveling on a straight line to a curve to a straight line. The curvature γ detected by the white line recognition sensor 2 is indicated by a solid line, and the gradient limit α is fixed to α0 by a two-dot chain line. , The corrected curvature p2 when the slope limit α is fixed to α2 by a one-dot chain line, and the corrected curvature γ ′ obtained by the curvature correction processing of the lane keep ECU 3 is indicated by a broken line. Show. When the gradient limit is fixed to α0, the corrected curvature p0 in the curve follows the change in the curve, but the correction curvature p0 also changes in the straight line following the detected curvature γ. On the other hand, when the gradient limit is fixed to α2, the corrected curvature p2 hardly changes in the straight line, but the corrected curvature p2 does not change in accordance with the change of the curve in the curve. When the gradient limitation is set in three stages according to the road shape, the corrected curvature γ ′ follows the change of the curve in the curve, and the corrected curvature γ ′ hardly changes in the straight line.

  The operation of the lane keeping device 1 will be described with reference to FIGS. 1 and 2. In particular, the curvature correction process in the lane keep ECU 3 will be described with reference to the flowchart of FIG. FIG. 4 is a flowchart showing the flow of curvature correction processing in the lane keep ECU of FIG. Here, a case where the lane keeping device 1 is activated by the driver will be described.

  The camera 2a captures an image of a road ahead of the vehicle and transmits data of the captured image to the white line recognition ECU 2b as an image signal PS. The white line recognition ECU 2b recognizes a pair of white lines that divide the lane from the captured image. Then, the white line recognition ECU 2b calculates the lane width, the lane center, the curve radius R and curvature γ, the yaw angle θ, and the vehicle offset D from the lane center from the pair of white lines. Further, the white line recognition ECU 2b transmits the information of the pair of white lines and each calculated information to the lane keep ECU 3 as an image signal GS at regular time intervals.

  At the time of activation, the lane keep ECU 3 sets α0 as the gradient limit α, resets the timers T0, T1, and T2 to 0, and sets 0 to the corrected curvature γ ′ and the previous corrected curvature γ ″ (S1). .

  The lane keep ECU 3 receives the image signal GS at regular intervals. Then, the lane keep ECU 3 acquires the curve radius R, the curvature γ, the yaw angle θ, and the offset D from the image signal GS (S2). The lane keep ECU 3 determines whether or not the curvature γ is larger than the curve determination threshold γ0 (S3). When it is determined in S3 that the curvature γ is equal to or less than the curve determination threshold γ0, the lane keep ECU 3 determines whether the curvature γ is larger than the curve determination threshold γ1 (S4).

  When it is determined in S3 that the curvature γ is larger than the curve determination threshold γ0, the lane keep ECU 3 increments the timer T0 and resets the timers T1 and T2 (S5). When it is determined in S4 that the curvature γ is larger than the curve determination threshold γ1, the lane keep ECU 3 increments the timer T1 and resets the timers T0 and T2 (S6). When it is determined in S4 that the curvature γ is equal to or less than the curve determination threshold γ1, the lane keep ECU 3 increments the timer T2 and resets the timers T0 and T1 (S7).

  The lane keep ECU 3 determines whether or not the timer T0 has exceeded the timer threshold τ (S8). When it is determined in S8 that the timer T0 has exceeded the timer threshold τ (that is, when the road shape is determined to be a curve having a relatively large curve radius), the lane keep ECU 3 sets α0 as the slope limit α (S9). ).

  When the timer T0 determines that the timer T0 is equal to or less than the timer threshold τ in S8, the lane keep ECU 3 determines whether the timer T1 exceeds the timer threshold τ (S10). When it is determined in S10 that the timer T1 has exceeded the timer threshold value τ (that is, when the road shape is determined to be a curve having a relatively small curve radius), α1 is set as the slope limit α (S11).

  When it is determined in S10 that the timer T1 is equal to or smaller than the timer threshold τ, the lane keep ECU 3 determines whether the timer T2 exceeds the timer threshold τ (S12). When it is determined in S12 that the timer T2 has exceeded the timer threshold value τ (that is, when the road shape is determined to be a straight line or a substantially straight line), α2 is set as the slope limit α (S13). If the timer T2 determines that the timer threshold value τ is equal to or smaller than the timer threshold τ in S12 (that is, if the road shape is not fixed), the slope limit α is not set.

  Subsequently, the lane keep ECU 3 determines whether or not (curvature γ−previous corrected curvature γ ″) is larger than the gradient limit α (S14). (Curvature γ−previous corrected curvature γ ″ in S14) ) Is greater than the gradient limit α (that is, when the curvature is increasing and the amount of increase from the previous corrected curvature γ ″ is greater than the gradient limit α), the lane keep ECU 3 (the previous corrected curvature) γ ″ + gradient limit α) is set as the corrected curvature γ ′ (S15).

  When it is determined in S14 that (curvature γ−previous corrected curvature γ ″) is equal to or less than the gradient limit α, the lane keep ECU 3 determines whether (curvature γ−previous corrected curvature γ ″) is less than the −gradient limit α. It is determined whether or not (S16). When it is determined in S16 that (curvature γ−previous corrected curvature γ ″) is smaller than −gradient limit α (that is, the amount of decrease from the previous corrected curvature γ ″ is smaller than the gradient limit α while the curvature is decreasing). If it is larger, the lane keep ECU 3 sets (previous corrected curvature γ ″ −gradient limit α) as the corrected curvature γ ′ (S17).

  When it is determined in S16 that (curvature γ−previous corrected curvature γ ″) is equal to or greater than the −gradient limit α (that is, when the amount of change from the previous corrected curvature γ ″ is equal to or smaller than the slope limit α), the lane keep The ECU 3 sets the detected curvature γ as a corrected curvature γ ′ (S18). Then, the lane keep ECU 3 sets the corrected curvature γ ′ as the previous corrected curvature γ ″ (S19), returns to S2, and waits until the next processing starts.

  The lane keep ECU 3 multiplies the corrected curvature γ ′, yaw angle θ, and offset D by gains G1, G2, and G3, respectively, and sets a target steering torque based on the multiplied values. The lane keep ECU 3 transmits a steering torque signal TS indicating the set target steering torque to the EPS ECU 10a.

  The EPS ECU 10a receives the steering torque signal TS, and multiplies the target steering torque indicated by the steering torque signal TS by a predetermined coefficient. Then, the EPS ECU 10a adds the multiplication value (additional torque) to the assist torque, and drives and controls the EPS motor 10b according to the assist torque + addition torque. The EPS motor 10b generates a predetermined torque under the control of the ECPECU 10a, and adds the torque to the steering mechanism. Then, an assist torque corresponding to the steering torque by the driver is applied to the steering mechanism, and an auxiliary additional torque for causing the vehicle to travel along the vehicle center is applied.

  According to this lane keeping device 1, by setting the gradient limit in three stages according to the shape of the road that has traveled in the past, it is possible to allow a large change in curvature as the curve radius decreases, and a change in curvature as the curve radius increases. Can be suppressed. Therefore, the corrected curvature can follow the actual curvature during the curve traveling, and noise included in the curvature detected from the corrected curvature can be removed during the straight traveling. As a result, the curvature of the traveling road can be obtained with high accuracy, and the accuracy of the lane keeping can be improved by using this curvature. Further, in the lane keeping device 1, since the road shape is fixed when the road shape is determined when the section of each curvature is traveled for a certain time or more, the road shape can also be determined with high accuracy.

  As mentioned above, although embodiment which concerns on this invention was described, this invention is implemented in various forms, without being limited to the said embodiment.

  For example, although the present embodiment is applied to the lane keeping device, it can be applied to the road curvature detection device itself, and can also be applied to other devices that use the curvature of the road, for example, automatic steering. Applicable to equipment.

  In this embodiment, the steering torque is applied by using the electric power steering device. However, the lane keeping device may be provided with an actuator that assists the rack and pinion. Therefore, the present invention can be applied to a vehicle that does not include a power steering device. Further, the present invention can be applied not only to an electric power steering device but also to a hydraulic power steering device, and may be configured to add steering torque by controlling an actuator that adjusts hydraulic pressure.

  In the present embodiment, the lane is detected by recognizing a pair of white lines. However, the lane may be detected by recognizing other lines such as a yellow line other than the white line, or a road shoulder. Alternatively, the lane may be detected by another method such as detecting the lane by recognizing a block that partitions the lane and the sidewalk. Further, although the present embodiment is applied to a road having a lane, the present invention can also be applied to a road having no lane. In this case, it is necessary to detect the shoulder of the road.

  Further, in this embodiment, the three-stage gradient limit is set according to the curvature of the traveling road that has traveled in the past. However, the two-stage gradient limit or the four-stage gradient limit or more may be set. Good.

It is a block diagram of the lane keeping apparatus which concerns on this Embodiment. It is a control block diagram of the lane keeping device according to the present embodiment. It is a figure which shows the relationship between a travel distance and the curvature of a travel path. It is a flowchart which shows the flow of the curvature correction process in lane keep ECU of FIG.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Lane keeping device, 2 ... White line recognition sensor, 2a ... Camera, 2b ... White line recognition ECU, 3 ... Lane keep ECU, 10 ... Electric power steering device, 10a ... EPSECU, 10b ... EPS motor

Claims (1)

In the travel path detection device that images the travel path by the imaging means and obtains the curvature of the travel path,
Obtain the curvature of the road at the position before the current, and allow the change in the curvature of the road when the curvature of the road at the position before the current is large compared to when it is small, and change of the allowable curvature A traveling path detection device for obtaining a curvature of a traveling path at a current position based on

Images