JP2002132343A - Travel control method for autonomous travel body - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a travel control method for an autonomous travel body which can obtain high tracking performance for a travel path by increasing the stable margin of a control system. SOLUTION: An image picked up by an image pickup device 2 mounted on the autonomous travel body 1 is divided into multiple areas 6 and feature quantities in the divided areas found by the divided areas 6 are compared with a specific threshold to extract a tracking object area 7; and a point P of interest is set in the tracking object area 7, the angle θ of steering of the autonomous travel body 1 for traveling to the set point P of interest is calculated, and the autonomous travel body 1 is steered and controlled with the steering angle θ.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a traveling control of an autonomous traveling body which autonomously travels on a place where a road boundary does not clearly exist, such as an uneven terrain and an unpaved road, using an image taken by an imaging device. About the method.

[0002]

2. Description of the Related Art A road such as an unpaved road without a clear boundary line is used for a vehicle such as a vehicle (vehive).
In the conventional technology for the cle) to follow, a method of directly obtaining the steering amount of the traveling body from the image of the road by a neural network which is a kind of signal conversion means is adopted. In this neural network, it is difficult to predict and analyze the problem because the logic of the traveling judgment is unknown, and every time there is a change in the setting position of the camera mounted on the traveling body or the traveling body itself, It is necessary to learn again, and there is a problem that the setting operation and the adjusting operation before the actual traveling are complicated.

Another conventional technique for following a road having a clear boundary line and a white line on the road is as follows.
This is disclosed in Japanese Patent No. 2,583,639. According to this conventional technique, a plurality of gazing points are set on an image obtained by capturing an image of a road ahead by an imaging device mounted on a traveling body, and a steering angle obtained by each gazing point information is determined by a lack of gazing point information such as an intersection. A method is disclosed in which a steering angle is obtained by weighting a vehicle, and a gazing point is used to steer a traveling body with the steering angle.

Further, another conventional technique similar to the conventional technique disclosed in Japanese Patent No. 2583639 is disclosed in Japanese Unexamined Patent Publication No.
No. 262207. According to this conventional technique, a driver specifies a linear marking such as a center line, a white line on a road shoulder, and a roadside zone drawn along a traveling road surface on an image tracking window, thereby tracking the linear marking. The operation is started, and the position of the image tracking window is moved within the display screen so as to be a predetermined position, so that the traveling direction of the traveling body can be controlled, and the advance detected by the obstacle detection window can be controlled. A method of controlling a traveling speed according to a distance to a vehicle is disclosed.

[0005]

The above-mentioned Patent No. 2584
In each of the conventional techniques disclosed in Japanese Patent Application Laid-Open No. 639 and Japanese Patent Application Laid-Open No. 4-262207, there is a problem that a control delay theoretically occurs due to the existence of an error and a stability margin of a control system is reduced. To calculate the steering amount from the error between the center of the image and the center of the road, PID (proportional in
Integral differential) calculation requires not only adjustment of the PID gain, but also the need to readjust the gain when changing the traveling vehicle, and it takes time to obtain the optimum gain. There is.

SUMMARY OF THE INVENTION It is an object of the present invention to increase the stability margin of a control system, eliminate the need for adjusting parameters such as a PID gain, and achieve a travel control of an autonomous traveling body capable of achieving high followability on a travel path. Is to provide a way.

[0007]

According to a first aspect of the present invention, an image captured by an imaging device mounted on a traveling body is divided into a plurality of regions, and the divided regions obtained for each divided region are obtained. Extract the tracking target area by comparing the feature amount within the predetermined threshold value, set a gazing point on this tracking target area, calculate the steering angle of the traveling body to head to this gazing point, A traveling control method for an autonomous traveling body, characterized in that the traveling body is subjected to steering control by the steering angle.

According to the present invention, an image captured by an imaging device is divided into a plurality of regions, a characteristic amount is obtained for each of the divided regions, and the characteristic amount is compared with a predetermined threshold value to thereby track the image. The target area is extracted.

A gazing point is set in the tracking target area extracted as described above, and the steering angle of the traveling body is calculated and obtained from the gazing point. The calculation of such a steering angle is disclosed in the aforementioned Japanese Patent No. 2583639 and Japanese Patent Laid-Open No. 4-26220.
Unlike each of the prior arts disclosed in Japanese Patent Application Publication No. 7-107, there is no assumption that there is an error, so that there is no control delay theoretically, and there is no problem that the stability margin of the control system is reduced.

[0010] Further, instead of calculating the steering amount from the error between the center of the image and the center of the road as in each of the above-mentioned prior arts,
Since a gazing point is set in a tracking target area composed of a plurality of divided areas, and a steering angle is obtained from the gazing point, it is not necessary to adjust parameters such as adjustment of a PID gain when the PID calculation is used. Also, even if various amounts involved in the calculation of the steering amount such as a change in the set position of the imaging device and a change in the vehicle are changed, it is not necessary to readjust the parameters, and many settings and adjustments before traveling are required. It is possible to obtain a high followability to the traveling path without requiring time and effort.

According to a second aspect of the present invention, as the feature quantity,
It is characterized in that the average value of the power spectrum in the divided area is used.

According to the present invention, an image captured by an imaging device is divided into a plurality of regions, and a power spectrum average value is calculated for each divided region. Even if the scales are different, the average value of the power spectrum becomes the same value, and a plurality of divided regions that do not include an error due to the scale change can be extracted as the tracking target regions.

According to a third aspect of the present invention, a travel route curve is obtained from the following target area by approximation, and the gazing point is set ahead of the travel route curve in the traveling direction.

According to the present invention, since the travel route curve is obtained by approximation from the following target area, there is no need for a white line or a road boundary line to exist on the actual travel road as in each of the prior arts. However, the place where the vehicle can travel is not limited to the place where the white line or the road boundary line exists. Further, since the traveling route curve is obtained by approximation, the influence of noise superimposed on the tracking target area such as a determination error of the tracking target area is small, and a highly accurate running curve with few errors can be obtained.

According to a fourth aspect of the present invention, the set position of the gazing point on the traveling route curve is changed according to at least one of a steering angle and a traveling speed of the traveling body. I do.

According to the present invention, the position of the gazing point on the traveling route curve can be changed by one or both of the steering angle and the traveling speed. By setting the distance close to the vehicle, it is possible to improve the followability with respect to the traveling route curve, and to obtain a high curve traveling property. In addition, when the vehicle is traveling straight, the gazing point is set far away from the traveling body, so that undesired changes in the steering angle can be reduced and high traveling stability can be obtained.

According to a fifth aspect of the present invention, at least one gazing point for determining a traveling speed command is set on a traveling route curve, and the traveling speed command is changed according to the curvature of the traveling path curve at the gazing point. It is characterized by the following.

According to the present invention, a gazing point for determining the traveling speed command is set independently of the gazing point for determining the steering angle, and the traveling command changes according to the curvature of the traveling path curve at the gazing point. Therefore, before entering a curved road, it is possible to reduce or increase the speed to the optimum speed at the point of gaze in the curved road, and to improve the followability to the traveling road and the traveling stability by adjusting the speed. it can.

According to a sixth aspect of the present invention, a spatial distance from a running object to a gazing point is obtained based on the coordinates of the gazing point on the image, and a traveling speed command is issued in accordance with the magnitude of the distance. It is characterized by changing.

According to the present invention, the traveling speed command is changed according to the spatial distance from the traveling object to the point of interest, so that traveling control of the traveling object such that the traveling object approaches or moves away from the point of interest becomes possible. For example, since the gazing point may be set on a moving following target, running control for following a human walking in front of the running body or another running object becomes possible.

According to a seventh aspect of the present invention, the gazing point is:
It is characterized in that it is set on a tracking target identified by a color characteristic from an image captured by an imaging device.

According to the present invention, the gazing point is set on the object to be tracked based on the color characteristics such as hue, saturation, and lightness. And it is possible to reliably follow and follow the following object.

According to the present invention, the traveling speed command is reduced as the spatial distance from the traveling body to the gazing point obtained from the coordinates of the gazing point on the image is reduced, and the spatial distance is smaller than a predetermined distance. In this case, the traveling control is performed by outputting a traveling stop command for setting the traveling speed to 0.

According to the present invention, if the pedestrian squats or falls while following the pedestrian in front of the vehicle, it is interpreted that the distance from the running body to the point of interest is short, Since the traveling body decelerates or stops, the following traveling control with high safety can be performed. Further, since the speed of the traveling body is controlled to be slower as the traveling body comes closer to the gazing point, the danger of collision between the following object and the traveling body is reduced, and safe following traveling is possible.

[0025]

FIG. 1 is a perspective view showing a traveling state of an autonomous traveling body 1 whose traveling is controlled by an autonomous traveling body traveling control method according to an embodiment of the present invention, and FIG. FIG. 3 is a diagram showing a state in which a traveling route curve and a gazing point P are set in a captured image 3 of an imaging device 2 mounted on a body 1.
The autonomous traveling body to which the traveling control method according to the present embodiment is applied is the autonomous traveling body 1 that travels autonomously on uneven terrain 4 to reach a destination and performs a predetermined operation according to, for example, a preset program. An image pickup device 2 is mounted on the autonomous traveling body 1, and a plurality of divided areas 6 are obtained from a color image 3 picked up by the image pickup device 2, that is, a hue image described later.
The tracking target area 7 composed of
A gazing point P is set above, a steering angle θ for heading toward the gazing point P is calculated, and the steering of the autonomous traveling body 1 is controlled by the steering angle θ.

As the image pickup device 2, a color CCD (charge coupled device) camera having high color reproducibility is used because unpaved roads and grassland are often mixed in an actual environment. A wide-angle lens is attached to this CCD camera so that the unpaved road does not deviate from the angle of view even when traveling on a curve. This wide-angle lens has an aperture function to cope with a large change in illuminance.

The autonomous vehicle 1 travels along an unpaved road by traveling in the center of the area 7 to be followed detected as an unpaved road from the image 3, and its algorithm will be described. First, a hue image is extracted from a color image captured by the imaging device 2. This hue is HLS
(Hue; Hue, lightness; Lightness, saturation; Saturation) The hue obtained by the conversion may be a hue obtained by performing a customization process that reflects a color distribution in the natural world. The HLS transformation means that a color image is represented by a coordinate system of R (red), G (green), and B (blue).
This is a coordinate conversion to a coordinate system represented by attributes such as brightness, hue, and saturation.

The hue image obtained in this way is divided into small divided areas 6, and the mean square value of the hue is calculated for each divided area 6. By averaging the squares, a power spectrum average value, which is a feature amount independent of a scale change due to a depth change, is obtained, and a discrimination ability higher than simply using the average value is obtained due to the spatial change in the hue image. be able to.

First, the hue range of an unpaved road, that is, the upper and lower limits of the hue, is measured in advance, and the mean square value of each divided area 6 is calculated as the square of the upper and lower limits. If the value is between the lower limit and the upper limit, the road is determined to be a road.

Here, the average value of the power spectrum described in the present case is an index indicating the magnitude of the spatial change in the image. If the average value of the power spectrum is Pav, each divided area 6 is composed of m × m pixels. When
Calculation of mean square value of image f (x, y) of divided area

[0031]

(Equation 1) Can be obtained as Regarding the power spectrum, see the literature ("Pattern Information Processing" by Makoto Nagao, Corona Co., Ltd., and "Handbook of Vibration Engineering" Taniguchi
Osamu Editor in Chief, Yokendo Co., Ltd.)

FIG. 3 is a flowchart for explaining a procedure for obtaining the steering amount θ of the autonomous traveling body 1, and FIG. 4 is a view showing a procedure for creating the entire traveling route 8 from the extracted tracking target area 7. 4 (1) shows a state where the tracking target area 7 is extracted, FIG. 4 (2) shows a state where the traveling route curve 9 is created by approximation, and FIG. 4 (4) shows the gazing point P on the image.
It is a figure which shows typically operation which transforms into space coordinate.

First, in step a1, the operation of extracting the steering amount θ of the autonomous vehicle 1 is started, and in step a2, the HLS conversion is performed from the color image obtained by the imaging device 2 as described above. To obtain a hue image that has been subjected to customization processing in which the hue obtained by reflecting the color distribution in the natural world is obtained. This hue image is divided into a plurality of divided regions 6, and the mean square value of the hue is calculated for each region. Further, in order to determine the tracking target area 7, a hue range corresponding to the tracking target area 7 is measured and determined in advance. This hue range is determined by setting the upper and lower limits of the hue. By extracting such a hue range, the tracking target area 7 including the plurality of divided areas 6 can be set as shown in FIG.

The tracking target area 7 set as described above is subjected to a labeling process since a plurality of tracking target areas may be detected in the image obtained as a result of the property recognition by the image processing. Then, only the tracking target area 7 to be traveled is selected from the labels. In the selected tracking target area 7, the label that is most frequently present in the lower center area of the image is selected as the road.

Next, in step a3, the tracking target area 7 of the image is divided in the vertical direction, and the center of gravity G
, And an approximate curve 8 connecting the respective centers of gravity G is obtained by the least square method. The approximation curve 8 cannot be sufficiently approximated when the degree is lower than the second order, and is approximated to the noise component when the degree is higher than the seventh order.

As described above, the travel route curve is obtained by approximation from the tracking target area. Therefore, unlike the prior arts, there is no need for a white line or a road boundary line to exist on the actual travel road. A possible location is not limited to a location where the white line or the road boundary exists. In addition, since the traveling route curve is obtained by approximation, the influence of noise superimposed on the tracking target area is small, and a highly accurate traveling curve with few errors can be obtained.

At step a4, the gazing point P is set on the approximate curve 8 representing the traveling route curve as described above as follows. First, y in the space up to the point of interest P
Horizontal distance on the axis (hereinafter referred to as the gazing point distance) L = y
₀ is set in advance to, j on the image corresponding to the fixation point distance L = y ₀ from the relative positional relationship between the actual ground the imaging apparatus
Find the coordinates (0, j ₀ ) on the axis.

It is assumed that the y-axis in the space is in the traveling direction of the traveling body, and the x-axis is in the direction perpendicular to the y-axis. Also, the j-axis on the image is taken in the vertical direction of the image, and the i-axis is taken in the horizontal direction.

Next, the coordinates (i ₀ , j ₀ ) of the gazing point P on the image are obtained by using the above-mentioned approximate curve 8.

Thereafter, in step a5, the coordinates (i ₀ ,
j ₀ ) and the spatial coordinates (x ₀ , y) of the gazing point P projected onto the ground
₀ ).

In step a6, a steering angle θ for heading toward the spatial coordinates of the gazing point P is calculated using an equation relating to steering of the autonomous vehicle 1 and an equation of motion of the autonomous vehicle 1. The calculation of the steering angle θ will be specifically described below.

FIG. 5 is a diagram showing the relationship between the gazing point P and the steering angle θ. The spatial coordinates of the gazing point P on the traveling route curve 8 obtained as described above are (x ₀ , y ₀ ), the distance in the axle direction from the center of the front wheel to the center of the rear wheel of the autonomous vehicle 1 is H, When the coordinates of the intersection with the front wheel center are (0, 0) and rear wheel steering is performed, and when the imaging device is installed above (0, 0), the steering angle θ for heading to the gazing point P is given by the following equation. , Θ = tan {2Hx ₀ / (x ₀ ² + y ₀ ² )} ⁻¹ (1)

If the gazing point distance L (L = y ₀ ) is increased, the vehicle goes to a distant point on the traveling route curve, so that a short path is expected where the road is curved. . In addition, when the gazing point distance L is reduced, the vehicle goes to a nearby point on the travel route curve, so that the followability to the curve is improved. On the other hand, the steering of the noise component generated when extracting the follow target is performed. And there is a disadvantage that predictive curve running performance cannot be obtained.

Therefore, in the present embodiment, the set position of the gazing point P on the traveling route curve is changed according to at least one of the steering angle θ and the traveling speed of the autonomous traveling body 1. Thereby, the position of the gazing point on the traveling route curve 9 can be changed by one or both of the steering angle θ and the traveling speed. Thus, the ability to follow the traveling route curve 9 is improved, and a high curved traveling property can be obtained. When the vehicle is traveling straight, the gazing point P is set far away from the autonomous vehicle 1 in the traveling direction, so that unstable changes in the steering angle θ due to the noise component are reduced, and high linear traveling performance is obtained. be able to.

FIG. 6 is a graph showing the relationship between the position of the gazing point P of the autonomous traveling body 1 and the steering angle θ for heading toward the gazing point P. As an example, when the driving speed command is 1 m / s and the autonomous driving gazing point distance L is 1 m, the gazing point P slightly changes from left to right from a state near the origin, which is also referred to as a large steering change. It can be seen that an error occurs. That is, such characteristics cause a large change in the steering angle due to an error in the gazing point P during straight running.

It can be seen that in order to improve the stability of the steering angle θ during straight running, the gazing point distance L should be increased until the change in the steering angle θ near the center becomes sufficiently gentle. Further, in the present embodiment, the maximum steering angle θ is
Since 30 ° is selected, the gazing point distance L that requires a steering angle θ greater than this is meaningless. Therefore, the steering angle θ is 30 °
When the gazing point distance L was obtained, it was about 2.8 m. From the above examination, the point of regard distance L is set to “2.8.
It is preferable to use three types of “m fixed”, “1.4 m fixed”, and “1.4 to 2.8 m variable”. “Fixed at 2.8 m” means that the gazing point distance L is kept constant at 2.8 m. “Fixed at 1.4” means that the gazing point distance L is 1.4.
m, and "variable from 1.4 to 2.8 m" means changing the gazing point distance L between 1.4 m and 2.8 m.

FIG. 7 is a graph showing the relationship between the gazing point distance L and the average steering angle θ when the gazing point distance L is made variable. The setting condition “fixed at 1.4 m” of the fixation point distance L
For comparison with "fixed at 2.8 m", the distance is adopted as a half distance of 2.8 m. In addition, "1.4-2.8m
As shown in FIG. 7, "variable" refers to a case where the gazing point distance L is increased during straight running and the gazing point distance L is reduced during curve running.

FIG. 8 is a graph showing the result of comparison of the running trajectories at the respective gazing point distances. The fixation point distance L is 1.4
The traveling locus of m (indicated by a dashed line in the figure) is near the start (upper left in the figure), and is different from the other locus because the soil color around this area has changed greatly, Is not extracted as a tracking target area. A running locus having a gazing point distance L of 2.8 m (shown by a broken line in the figure) has a slightly lower curve following ability than other running locus. This is because when the gazing point distance is large, a short path is provided. Gaze point distance L
Is variable to 1.4 to 2.8 m (shown by a solid line in the figure).
It is almost the same as the traveling locus of m, and has good followability during a curve. In the vicinity of the start, the running locus is almost the same as the 2.8 m running locus, indicating that the vehicle is not affected by disturbance.

FIG. 9 is a graph showing the results of capturing changes in the steering angle (actually measured values) at each point of regard. When the gazing point distance L is 2.8 m, the steering angle θ is as shown in FIG.
As shown in the figure, only a gradual change for road following is seen. Further, as shown in FIG. 9 (2), the steering angle θ when the gazing point distance L is 1.4 m has a very large variation other than the variation for following the road. Further, the steering angle θ in the case where the gazing point distance L is made variable from 1.4 to 2.8 m has the same gradual change as in the case of 2.8 m, though a slight change is added.

Regarding the control cycle, the unpaved road corresponding to the section W from the start point 0 m to about 120 m is a straight line. 8m and 1.4
The fluctuation of the steering angle θ when variable to ２2.8 m is much smaller than the case of the gazing point distance L1.4 m (fluctuation of about one-third peak-to-peak) and can be said to be stable. .

From the above results, the characteristics relating to the running locus and the steering angle can be confirmed, and it can be seen that the running characteristics are the best when the gazing point distance L is varied from 1.4 to 2.8 m.

As described above, when the setting condition of the gazing point distance L is "fixed to 1.4 m", the followability to the road when traveling on a curve is good as shown by the dashed line, but the steering when straight ahead is performed. Poor angular stability. In the case of "fixed at 2.8 m", as shown by the broken line, the followability to the road at the time of the curve is inferior, but the steering angle stability at the time of going straight is good. Further, in the case of "1.4 to 2.8 m variable", as shown by the solid line, both the followability to the road when traveling on a curve and the steering angle stability when traveling straight ahead are good.

FIG. 10 is a graph for explaining a traveling control method using variable speed traveling according to another embodiment of the present invention. FIG. 10 (1) shows the variation of the steering angle θ and the speed command Ss.
10 (2) shows the relationship between the differential value of the travel route curve at the point of gaze P and the speed command Sd. When traveling at a constant high speed, for example, at a speed exceeding the limit speed of 1 m / s to 2 m / s of the autonomous vehicle used in this experiment, if the vehicle is decelerated after entering a curved road, the deceleration becomes insufficient. Get out of the course on a curved road. As a material for determining a curved road, the traveling control is performed in consideration of the degree of turning ahead of the road.

More specifically, a second gazing point P 'for speed control, which is different from the gazing point P for steering control, is set at a relatively distant point on the traveling route curve. Is calculated and obtained, and this is adopted as a parameter of the degree of bending ahead of the road.

As shown in FIG. 10 (1), a steering angle θ for heading toward the gazing point P for steering control is detected as a current steering angle, and a speed command Ss corresponding to the current steering angle is obtained. As shown in FIG. 10 (2), the differential value of the traveling route curve at the point of interest P 'for speed control is detected as the degree of the curve ahead of the road, and the speed command Sd corresponding to the differential value is obtained. Only the smaller one of the speed commands Ss and Sd obtained as described above is used so that the vehicle travels at a high speed only when the currently traveling road is a straight road and the road ahead is also a straight road. Is selected as the traveling speed command and output to the drive system of the traveling body 1.

In this way, the vehicle travels at a high speed on a straight road,
On a curved road, by controlling the speed to be variable so as to run at a low speed in accordance with the curvature, it is possible to realize a higher-speed following travel.

As still another embodiment of the present invention, in addition to the traveling control of the embodiment shown in FIG. 10, in order to prevent a single-shot acceleration on a straight road, the average value of past speed commands is used. Compare with. In other words, using the three pieces of information of the current steering value, the degree of turning ahead of the road, and the average value of the past speed commands, the current traveling part is a straight road, and the front of the road is a straight road. The control is performed such that the value of the speed command is increased so as to satisfy the condition that the straight road is not one-shot but continuous. Thus, the problem of acceleration during traveling on a straight road is eliminated, and traveling stability on a straight road can be improved.

In another embodiment of the present invention, the gazing point P
Is set on a leader who is a tracking target identified by color characteristics from an image captured by the imaging device 2.

When the following object is a leader who walks in front of the autonomous vehicle 1 in the traveling direction, it is necessary to know the distance and azimuth to this leader, or the spatial position. Therefore, the distance and the azimuth to the leader are estimated from the result of extracting the color of the clothes as the marker of the leader.

In such a human-following traveling, a function of following the vehicle while keeping the distance between the leader and the autonomous traveling body 1 substantially constant is required. Therefore, estimation of the distance is particularly important. Also, the movement of the leader, for example, when viewed from the front, when viewed from the side, and when falling, there is a change in the appearance, and if highlights due to specular reflection of sunlight occur on clothes, The color of the highlight changes. For this reason, the size of the clothing image extracted using the color characteristics changes, and it is necessary to perform distance estimation that is not affected by the size. From such a viewpoint, a method of estimating the distance to the leader from the vertical coordinates of the clothing image in the image may be employed.

According to such a configuration, the point of gaze P is displayed on the leader by the color characteristics such as the hue and saturation of the clothing of the leader.
Is set, the discriminating ability is high, and it is possible to reliably follow and guide the leader.

In still another embodiment of the present invention, the smaller the spatial distance from the traveling object 1 to the gazing point P obtained from the coordinates of the gazing point P on the image, the smaller the traveling speed command is. If the distance is smaller than the predetermined distance, a travel stop command for setting the travel speed to 0 is output to control the travel.

According to such a configuration, when the vehicle is following the pedestrian in front of the vehicle, the pedestrian can squat,
When the vehicle falls, it is interpreted that the distance from the vehicle to the gazing point P is approached, and the vehicle decelerates or stops, so that the follow-up traveling control with high safety can be performed. Further, since the speed of the traveling body is controlled to be slower as the traveling body comes closer to the gazing point, the danger of collision between the following object and the traveling body is reduced, and safe following traveling is possible.

FIG. 11 is a diagram for explaining an algorithm for extracting a tracking object according to still another embodiment of the present invention. In this embodiment, the tracking target area is basically detected using the root mean square value (power spectrum) of the hue. However, in order to reduce clutter, threshold value processing of saturation is added. Accordingly, even a red-based hue having low saturation is removed as clutter, and a labeling process is performed to extract a label in the image as a candidate for the plurality of pylons 11 to be followed. Each pylon 11
Is a sign made of synthetic resin in the shape of a hollow triangular pyramid, which is used for road construction and the like and is colored in a relatively conspicuous color such as red.

Of the plurality of pylons 11, the second pylon from the bottom is extracted on the image, and the point of interest P1 is shifted to the point shifted horizontally by a predetermined distance from the center of gravity of the pylon on the image. The autonomous vehicle 1 is caused to follow the pylon 11 while sequentially changing the setting in accordance with the position change in the image of the pylon 11 due to the movement of the vehicle. Can be.

[0066]

According to the first aspect of the present invention, an image picked up by an image pickup device is divided into a plurality of regions, and a feature amount in each divided region is calculated for each divided region, and a predetermined threshold is calculated. The tracking target area is extracted by comparing the value with the value. A gazing point is set in the tracking target area extracted in this manner, and the steering angle of the vehicle is obtained from the gazing point, so that there is no control delay and the problem that the stability margin of the control system is reduced can be eliminated.

Further, since a gazing point is set in a tracking target area composed of a plurality of divided areas, and a steering angle is obtained from the gazing point, it is involved in calculation of a steering amount such as a change in a set position of an imaging device and a change in a vehicle. Even if the various amounts change, it is not necessary to adjust the parameters, and it is possible to obtain high followability with respect to the traveling path without requiring much time and labor for the setting operation and the adjusting operation before traveling.

According to the second aspect of the present invention, the image taken by the imaging device is divided into a plurality of regions, and the average value of the power spectrum is calculated for each of the divided regions to include an error due to a scale change. A plurality of undivided regions can be extracted as tracking target regions.

According to the third aspect of the present invention, the travel route curve is obtained by approximation from the tracking target area.
Unlike the prior arts described above, there is no need for a white line or a road boundary line to exist on the actual traveling road, and a place where the vehicle can travel is not limited to a place where the white line or the road boundary line exists. Further, since the traveling route curve is obtained by approximation, it is possible to process the noise component superimposed on the tracking target area on the approximation calculation, and it is possible to obtain a highly accurate traveling curve with few errors.

According to the fourth aspect of the present invention, the position of the gazing point on the traveling route curve can be changed by one or both of the steering angle and the traveling speed. Is set close to the vehicle, the followability to the traveling route curve is improved, and a high curve traveling property can be obtained. In addition, by setting the gazing point to be far away from the vehicle during straight running, undesired changes in the steering angle can be reduced, and high straight running performance can be obtained.

According to the present invention, a gazing point for determining the traveling speed command is set independently of the gazing point for determining the steering angle, and the gazing point is determined according to the curvature of the traveling path curve at the gazing point. Since the traveling command changes, before entering the curved road, it is possible to reduce or increase the speed to the optimum speed at the gazing point on the curved road. Can be improved.

According to the sixth aspect of the present invention, the traveling speed command is changed according to the spatial distance from the vehicle to the gazing point, so that the traveling body moves toward or away from the gazing point. Control becomes possible. For example,
Since the gazing point may be set on the moving following object, traveling control for following a human or other traveling object walking in front of the traveling body becomes possible.

According to the present invention, since the gazing point is set on the object to be tracked based on color characteristics such as hue and saturation, the discriminating ability is high, and guidance is performed by reliably following the object to be tracked. It is possible to do.

According to the present invention, when the pedestrian squats or falls while following the pedestrian in front of the vehicle, the distance from the running body to the point of interest is underground. Since the traveling body is decelerated or suddenly stopped, the following traveling control with high safety can be performed.
Further, since the speed of the traveling body is controlled so as to be slower as the point of interest approaches the traveling body, the danger of collision between the following object and the traveling body is reduced, and safe following traveling becomes possible.

[Brief description of the drawings]

FIG. 1 is a perspective view showing a traveling state of an autonomous traveling body 1 whose traveling is controlled by a traveling control method for an autonomous traveling body according to an embodiment of the present invention.

FIG. 2 is a diagram showing a state in which a tracking target area is extracted from an image obtained by an image captured by an imaging device mounted on an autonomous vehicle, and a traveling route curve and a gazing point are set on the area; is there.

FIG. 3 is a flowchart for explaining a procedure for obtaining a steering amount θ of the autonomous traveling body 1.

FIG. 4 is a diagram showing a procedure for creating a traveling route 8 from the extracted tracking target area 7, and FIG.
Is extracted, FIG. 4 (2) shows a state in which the travel route curve 9 is created by approximation, and FIG. 4 (3) shows a state in which the gazing point P is set on the travel route curve. FIG.
FIG. 4D is a diagram schematically illustrating an operation of converting the gazing point P on the image into spatial coordinates.

FIG. 5 is a diagram showing a relationship between a gazing point P and a steering angle θ.

FIG. 6 shows the position of the gazing point P of the autonomous vehicle 1 and the gazing point P;
6 is a graph showing a relationship with a steering angle θ for heading toward.

FIG. 7 is a gazing point distance L when the gazing point distance L is variable.
5 is a graph showing a relationship between the average steering angle θ and the average steering angle θ.

FIG. 8 is a diagram for explaining an algorithm for extracting a tracking target according to still another embodiment of the present invention.

FIG. 9 is a graph showing a result of capturing a change in a steering angle (actually measured value) at each gazing point distance.

FIG. 10 is a graph for explaining a traveling control method using variable speed traveling according to another embodiment of the present invention.
(1) shows the relationship between the change amount of the steering angle θ and the speed command Ss, and FIG. 10 (2) shows the relationship between the differential value of the travel route curve at the point of regard P and the speed command Sd.

FIG. 11 is a diagram for explaining an algorithm for extracting a tracking target according to still another embodiment of the present invention.

[Explanation of symbols]

 Reference Signs List 1 autonomous vehicle 2 imaging device 3 extracted image of tracking target area 4 rough terrain 6 divided area 7 tracking target area θ steering angle L fixation distance

────────────────────────────────────────────────── ───

[Procedure amendment]

[Submission date] December 15, 2000 (200.12.
15)

[Procedure amendment 1]

[Document name to be amended] Statement

[Correction target item name] 0018

[Correction method] Change

[Correction contents]

According to the present invention, the gazing point for determining the traveling speed command is set independently of the gazing point for determining the steering angle, and the traveling command changes according to the curvature of the traveling path curve at the gazing point. Therefore, before entering a curved road, it is possible to reduce or increase the speed to the optimum speed at the point of gaze in the curved road, and to improve the followability to the traveling road and the traveling stability by adjusting the speed. it can.

[Procedure amendment 2]

[Document name to be amended] Statement

[Correction target item name] 0025

[Correction method] Change

[Correction contents]

[0025]

FIG. 1 is a perspective view showing a traveling state of an autonomous traveling body 1 whose traveling is controlled by an autonomous traveling body traveling control method according to an embodiment of the present invention, and FIG. FIG. 3 is a diagram showing a state in which a traveling route curve and a gazing point P are set in a captured image 3 of an imaging device 2 mounted on a body 1.
The autonomous traveling body to which the traveling control method according to the present embodiment is applied is the autonomous traveling body 1 that travels autonomously on uneven terrain 4 to reach a destination and performs a predetermined operation according to, for example, a preset program. An image pickup device 2 is mounted on the autonomous traveling body 1, and a plurality of divided areas 6 are obtained from a color image 3 picked up by the image pickup device 2, that is, a hue image described later.
The tracking target area 7 composed of
A gazing point P is set above, a steering angle θ for heading toward the gazing point P is calculated, and the steering of the autonomous traveling body 1 is controlled by the steering angle θ. The “following” described in the present invention refers to an act of moving along a route such as a road or an act of following an object moving ahead.

[Procedure amendment 3]

[Document name to be amended] Statement

[Correction target item name] 0037

[Correction method] Change

[Correction contents]

At step a4, the point of interest P is set on the approximate curve 8 representing the travel route curve as described above as follows. First, the horizontal distance on the y-axis in space to the point of interest P (hereinafter, referred to as the point of interest) L = y ₀
Are set in advance, and the coordinates (0, j ₀ ) on the j-axis on the image corresponding to the gazing point distance L = y ₀ are obtained from the relative positional relationship between the imaging device and the actual ground.

[Procedure amendment 4]

[Document name to be amended] Statement

[Correction target item name] 0042

[Correction method] Change

[Correction contents]

FIG. 5 is a diagram showing the relationship between the gazing point P and the steering angle θ. The spatial coordinates of the gazing point P on the traveling route curve 8 obtained as described above are (x ₀ , y ₀ ), the distance in the axle direction from the center of the front wheel to the center of the rear wheel of the autonomous vehicle 1 is H, When the coordinates of the intersection with the center of the front wheel are (0, 0) and the rear wheel is to be steered, the steering angle θ for heading toward the gazing point P is as follows: θ = tan ⁻¹ ｛2Hx ₀ / (x ₀ ² + y ₀₎ ² )｝ …… (1)

[Procedure amendment 5]

[Document name to be amended] Statement

[Correction target item name] 0043

[Correction method] Change

[Correction contents]

If the gazing point distance L (L = y ₀ ) is increased, the vehicle goes to a distant point on the traveling route curve, so that a short path is expected where the road is curved. . Further, when the gazing point distance L is reduced, the vehicle goes to a nearby point on the travel route curve, so that the followability to the curve is improved. However, the steering due to the noise component generated when the target to be followed is extracted is improved. There is an inconvenience that the influence on the corner is large and a predictable curve running performance cannot be obtained.

[Procedure amendment 6]

[Document name to be amended] Statement

[Correction target item name] 0045

[Correction method] Change

[Correction contents]

FIG. 6 is a graph showing the relationship between the position of the gazing point P of the autonomous traveling body 1 and the steering angle θ for heading toward the gazing point P. As an example, when the gazing point distance L is 1 m, it can be seen that a large change in the steering angle occurs when the gazing point P slightly changes from left to right from a state near the origin. That is, such characteristics cause a large change in the steering angle due to an error in the gazing point P during straight running.

[Procedure amendment 7]

[Document name to be amended] Statement

[Correction target item name] 0046

[Correction method] Change

[Correction contents]

It can be seen that in order to improve the stability of the steering angle θ during straight running, the gazing point distance L should be increased until the change in the steering angle θ near the center becomes sufficiently gentle. Further, in the present embodiment, the maximum steering angle θ is
Since 30 ° is selected, the gazing point distance L that requires a steering angle θ greater than this is meaningless. Therefore, the steering angle θ is 30 °
When the gazing point distance L was obtained, it was about 2.8 m. From the above examination, the point of regard distance L is set to “2.8.
It is preferable to use three types of “m fixed”, “1.4 m fixed”, and “1.4 to 2.8 m variable”. “Fixed at 2.8 m” means that the gazing point distance L is kept constant at 2.8 m. “Fixed at 1.4” means that the gazing point distance L is 1.4.
m, and "variable from 1.4 to 2.8 m" means changing the gazing point distance L between 1.4 m and 2.8 m. The condition for setting the gazing point distance L “1.
"4 m fixed" is used for comparison with "2.8 m fixed".
This is adopted as half the distance of 8 m.

[Procedure amendment 8]

[Document name to be amended] Statement

[Correction target item name] 0047

[Correction method] Change

[Correction contents]

FIG. 7 is a graph showing the relationship between the gazing point distance L and the average steering angle θ when the gazing point distance L is made variable. As shown in FIG. 7, "1.4 to 2.8 m variable" refers to a case where the gazing point distance L is increased during straight traveling and the gazing point distance L is decreased during curve running.

[Procedure amendment 9]

[Document name to be amended] Statement

[Correction target item name] 0049

[Correction method] Change

[Correction contents]

FIG. 9 is a graph showing the results of capturing changes in the steering angle (actually measured values) at each fixation point distance. When the gazing point distance L is 2.8 m, as shown in FIG. 9A, only a gradual change in the steering angle θ for following the road is observed. The steering angle θ when the gazing point distance L is 1.4 m
As shown in FIG. 9 (2), a very large change occurs in addition to the change for following the road. Further, the steering angle θ in the case where the gazing point distance L is made variable from 1.4 to 2.8 m has the same gradual change as in the case of 2.8 m, though a slight change is added.

[Procedure amendment 10]

[Document name to be amended] Statement

[Correction target item name] 0050

[Correction method] Change

[Correction contents]

Regarding the control cycle, since the unpaved road corresponding to the section W from the start point 0 m to about 120 m is a straight line, the stability of the steering angle when traveling straight ahead is compared. 0.8m and 1.4 ~
The fluctuation of the steering angle θ when the variable angle is set to 2.8 m is much smaller than the case of the gazing point distance L1.4 m (a fluctuation of about one-third peak-to-peak) and can be said to be stable.

[Procedure amendment 11]

[Document name to be amended] Statement

[Correction target item name] 0054

[Correction method] Change

[Correction contents]

More specifically, a second gazing point P 'for speed control, which is different from the gazing point P for steering control, is set relatively distant on the traveling route curve, and the gazing point P' Is calculated and obtained, and this is adopted as a parameter of the degree of bending ahead of the road.

[Procedure amendment 12]

[Document name to be amended] Statement

[Correction target item name] 0055

[Correction method] Change

[Correction contents]

As shown in FIG. 10A, the steering angle θ for heading toward the gazing point P for steering control is detected as the current steering angle, and the speed command Ss corresponding to the current steering angle is obtained. As shown in FIG. 10 (2), the differential value of the traveling route curve at the point of interest P 'for speed control is detected as the degree of the curve ahead of the road, and the speed command Sd corresponding to the differential value is obtained. Each speed command S determined as described above so that the vehicle travels at a high speed only when the currently traveling road is a straight road and the road ahead is also a straight road.
The smaller one of s and Sd is selected as the traveling speed command and output to the drive system of the traveling body 1.

[Procedure amendment 13]

[Document name to be amended] Statement

[Correction target item name] 0057

[Correction method] Change

[Correction contents]

As still another embodiment of the present invention, in addition to the traveling control of the embodiment shown in FIG. 10, in order to prevent a single-shot acceleration on a straight road, the average value of past speed commands is used. Compare with. That is, using the three pieces of information of the current steering angle, the degree of turning ahead of the road, and the average value of the past speed commands, that the currently traveling portion is a straight road, and that the front of the road is a straight road, The control is performed such that the value of the speed command is increased so as to satisfy the condition that the straight road is not one-shot but continuous. This eliminates the problem of accelerating while driving on a short straight road,
The running stability on a straight road can be improved.

[Procedure amendment 14]

[Document name to be amended] Statement

[Correction target item name] 0058

[Correction method] Change

[Correction contents]

In another embodiment of the present invention, the gazing point P
Is set on the following target object identified by the color characteristic from the image captured by the imaging device 2.

[Procedure amendment 15]

[Document name to be amended] Statement

[Correction target item name] 0065

[Correction method] Change

[Correction contents]

Of the plurality of pylons 11, the second pylon from the bottom is extracted on the image, and the point of interest P1 is shifted to the point shifted horizontally by a predetermined distance from the center of gravity of the pylon on the image. The autonomous vehicle 1 is caused to follow the pylon 11 while sequentially changing the setting in accordance with the position change in the image of the pylon 11 due to the movement of the vehicle. Can be.
Further, in the above-described embodiment, a vehicle moving on land has been described as an autonomous vehicle, but the present invention is also applicable to a ship or submarine moving on water or underwater, or an aircraft flying in the air.

[Procedure amendment 16]

[Document name to be amended] Statement

[Correction target item name] 0069

[Correction method] Change

[Correction contents]

According to the third aspect of the present invention, the travel route curve is obtained by approximation from the tracking target area.
Unlike the prior arts described above, there is no need for a white line or a road boundary line to exist on the actual traveling road, and a place where the vehicle can travel is not limited to a place where the white line or the road boundary line exists. Further, since the traveling route curve is obtained by approximation, it is possible to process a noise component superimposed on the tracking target area, and it is possible to obtain a high-accuracy traveling curve with few errors.

[Procedure amendment 17]

[Document name to be amended] Statement

[Correction target item name] 0071

[Correction method] Change

[Correction contents]

According to the fifth aspect of the present invention, a gazing point for determining a traveling speed command is set independently of a gazing point for determining a steering angle, and the gazing point is determined according to the curvature of a traveling path curve at the gazing point. Since the traveling command changes, before entering the curved road, it is possible to reduce or increase the speed to the optimum speed at the gazing point on the curved road. Can be improved.

[Procedure amendment 18]

[Document name to be amended] Statement

[Correction target item name]

[Correction method] Change

[Correction contents]

According to the present invention, when the pedestrian squats or falls while following the pedestrian in front of the vehicle, the distance from the traveling body to the point of interest decreases. As a result, the traveling body decelerates or stops, so that it is possible to perform the following traveling control with high safety. Further, since the speed of the traveling body is controlled to be slower as the traveling body comes closer to the gazing point, the danger of collision between the following object and the traveling body is reduced, and safe following traveling is possible.

[Procedure amendment 19]

[Document name to be amended] Statement

[Correction target item name] Fig. 8

[Correction method] Change

[Correction contents]

FIG. 8 is a graph showing a result obtained by capturing a change in a running locus (actually measured value) at each fixation point distance.

[Procedure amendment 20]

[Document name to be amended] Statement

[Correction target item name] Fig. 9

[Correction method] Change

[Correction contents]

FIG. 9 is a graph showing a result of capturing a change in a steering angle (actually measured value) at each gazing point distance.

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) B62D 113: 00 B62D 113: 00 (72) Inventor Yasuyuki Kitamura 1 Kawasaki-cho, Kakamigahara-shi, Gifu Kawasaki Heavy Industries, Ltd. F term in Gifu factory (reference) 3D032 CC20 DA04 DA23 DA81 DC04 DC38 EB04 EC34 GG07 5B057 AA16 BA02 DA08 DB02 DC05 5C054 AA02 CC02 CD03 CE11 FC11 FC12 FF07 HA28 5H301 AA01 CC06 GG01 HH03

Claims (8)

[Claims] An image captured by an imaging device mounted on a traveling body is divided into a plurality of regions, and a feature amount in each divided region obtained for each divided region is compared with a predetermined threshold value. Extracting a follow-up target area, setting a gazing point on the tracing-target area, calculating a steering angle of the traveling body toward the gazing point, and performing steering control of the traveling body based on the steering angle. A traveling control method for an autonomous traveling body. 2. The method according to claim 1, wherein an average value of the power spectrum in the divided area is used as the characteristic amount. 3. The autonomous vehicle according to claim 1, wherein a trajectory curve is obtained from the following target area by approximation, and the gazing point is set ahead of the traveling direction on the trajectory curve. Travel control method. 4. The autonomous system according to claim 3, wherein the set position of the gazing point on the traveling route curve is changed according to at least one of a steering angle and a traveling speed of the traveling body. A traveling control method for the traveling body. 5. The method according to claim 1, wherein one or more gazing points for determining a traveling speed command are set on the traveling path curve, and the traveling speed command is changed according to the curvature of the traveling path curve at the gazing point. 5. The travel control method for an autonomous traveling body according to 3 or 4. 6. A method according to claim 1, wherein:
The travel control method for an autonomous traveling body according to claim 1 or 2, wherein a spatial distance from the traveling body to a point of interest is obtained, and a traveling speed command is changed according to the magnitude of the distance. 7. The autonomous mobile unit using a gazing point according to claim 6, wherein the gazing point is set on an object to be tracked identified by a color characteristic from an image captured by an imaging device. Travel control method. 8. The travel speed command is reduced as the spatial distance from the traveling body to the gazing point obtained from the coordinates of the gazing point on the image is reduced. The travel control method for an autonomous traveling body according to claim 6 or 7, wherein the travel control is performed by outputting a travel stop command.