JP2019012063A - Obstacle detection device for construction vehicle - Google Patents

Abstract

To provide an obstacle detection device for a construction vehicle superior in detection accuracy of an obstacle.SOLUTION: An obstacle detection device mounted on a construction vehicle includes: a TOF type distance image sensor configured to measure a distance on the basis of a time difference between projected light and reflected light; and a control unit configured to determine presence/absence of an obstacle on the basis of data measured by the distance image sensor. With respect to measurement data, when defining a component from an end of the vehicle in a vehicle traveling direction as x data; a component in a vehicle width direction as y data and a component in a height direction from the ground as z data, the control unit is configured to perform a height selection step of determining the presence/absence of the obstacle by selecting only the measurement data in which the z data is larger than a threshold value T1.SELECTED DRAWING: Figure 7

Description

  The present invention relates to an obstacle detection device for a construction vehicle.

  In the rolling roller, for example, when the rolling force is applied to the curb, the driver operates while paying attention to the rolling surface around the curb, so that attention to the traveling direction tends to be neglected. Therefore, accidents that come into contact with surrounding workers tend to occur particularly when the vehicle is moving backward.

  In response to this problem, there are known alarm devices that issue an alarm when a person or object is detected at a certain distance using radio waves or ultrasonic waves, or an automatic stop device that automatically stops the vehicle (for example, patents). References 1-3). Patent Document 1 discloses a magnetic field generator mounted on a vehicle, an IC tag attached to an operator, a detection device that detects a radio wave transmitted from the IC tag, and a vehicle when the detection device detects a radio wave. An emergency stop device comprising an engine stop device for stopping the engine is described. Patent Document 2 discloses a trigger signal output means mounted on a vehicle, an ID tag attached to an operator, a receiving unit that receives an ID number output from the ID tag, and a receiving unit that receives the ID number. Describes a stop system comprising stop means for stopping the vehicle. Patent Document 3 describes an ultrasonic obstacle detection device.

Japanese Patent Application Laid-Open No. 2006-153558 JP 2017-10484 A JP 2006-17496 A

  Since the techniques of Patent Documents 1 and 2 require the tag to be attached to the worker, there is a problem that the cost is likely to increase when there are a large number of workers, and the worker forgets to attach the tag. I can get up.

  In addition, when detecting obstacles with ultrasonic waves, etc., there is a risk of recognizing ground irregularities and slopes as obstacles. The problem of stopping occurs.

  Also, if the construction site is fog or rain, or if water vapor or dust etc. is floating, these may be recognized as obstacles. Problem that stops.

  The present invention was created to solve such problems, and an object of the present invention is to provide an obstacle detection device for a construction vehicle that is excellent in obstacle detection accuracy.

  In order to solve the above-mentioned problems, the present invention is an obstacle detection device mounted on a construction vehicle, wherein a distance image sensor of a TOF method that measures a distance from a time difference between projected light and reflected light, and the distance image sensor A control device for determining the presence or absence of an obstacle based on the measurement data of the vehicle, and for the measurement data, the vehicle traveling direction component from the vehicle end is x data, the vehicle width direction component is y data, and the height from the ground When the vertical component is z data, the control device performs a height selection step of selecting only measurement data whose z data is larger than a threshold value T1 and determining the presence or absence of an obstacle.

  According to the present invention, the accuracy of obstacle detection can be increased by using a distance image sensor of the TOF method. There is no need for workers to attach tags. And by performing a height selection step, the unevenness of the ground can be ignored, and useless obstacle detection can be reduced.

  Further, the present invention is an obstacle detection device mounted on a construction vehicle, a TOF type distance image sensor that measures a distance from a time difference between projected light and reflected light, measurement data of the distance image sensor, And a control device that determines the presence or absence of an obstacle based on the reflection intensity of the reflected light, and the control device selects only measurement data whose reflection intensity of the reflected light is greater than a threshold value T6 and obstructs the obstacle. A reflection intensity selecting step for determining the presence or absence of the image is performed.

  According to the present invention, the accuracy of obstacle detection can be increased by using a distance image sensor of the TOF method. There is no need for workers to attach tags. Further, the reflection intensity of reflected light from water vapor or dust is low compared to the reflection intensity of reflected light from a solid material. Therefore, by providing a control device that determines the presence or absence of obstacles based on the measurement data of the distance image sensor and the reflection intensity of reflected light, it is erroneously determined that fine particles such as water vapor and dust floating in the air are obstacles. Can be reduced. And the unevenness | corrugation of the ground can be disregarded and useless obstacle detection can be reduced.

  In the first aspect of the present invention, the control device calculates a slope of the distribution from the x data and the z data of the measurement data sorted in the height sorting step, and determines whether there is an obstacle based on the slope. A determination step is performed.

  According to the present invention, it is possible to reduce the risk of determining an uphill as an obstacle.

  In the first aspect of the present invention, the control device calculates a slope of the distribution from x data and z data of the measurement data selected in the reflection intensity selection step, and determines whether there is an obstacle based on the slope. A determination step is performed.

  According to the present invention, it is possible to reduce the risk of determining an uphill as an obstacle.

In the invention, it is preferable that the inclination is an inclination m calculated by a least square method in a scatter diagram of the x data on the horizontal axis and the z data on the vertical axis.
Further, the present invention is characterized in that it is determined that there is an obstacle when the slope m is larger than a threshold value T2, smaller than 0, or division by zero.

  According to the present invention, it is possible to further reduce the risk of determining an uphill as an obstacle.

Further, the present invention is characterized in that the inclination is an inclination a calculated by a least square method in a scatter diagram of the z data on the horizontal axis and the x data on the vertical axis.
Further, the present invention is characterized in that it is determined that there is an obstacle when the slope 1 / a obtained from the slope a is larger than a threshold value T4 or smaller than 0.
The present invention is characterized in that it is determined that there is an obstacle when the slope a is smaller than a threshold value T5.

  According to the present invention, it is possible to further reduce the risk of determining an uphill as an obstacle.

  In the present invention, the control device calculates a determination coefficient from x data and z data of the measurement data selected in the height selection step, and when it is determined that there is no obstacle in the first determination step. The second determination step of determining that there is an obstacle when the determination coefficient is smaller than the threshold value T3 is performed.

  According to the present invention, the obstacle detection accuracy is further improved.

  According to the present invention, an obstacle detection device with excellent obstacle detection accuracy is obtained.

It is explanatory drawing which shows the detection range of the obstacle detection apparatus with which the tire roller was mounted | worn, (a), (b) is a top view and a side view, respectively. (A) is a side view of a detection range when a detection target is an operator, and (b) is a graph of calculated x data and z data. (A) is a side view of a detection range when a detection target is a box with a height, and (b) is a graph of calculated x data and z data. (A) is a side view of a detection range when a detection target is a box with a low height, and (b) is a graph of calculated x data and z data. (A) is a side view of a detection range when a detection target is a slope having a relatively slope, and (b) is a graph of calculated x data and z data. (A) is a side view of a detection range when a detection target is a slope with a relatively gentle slope, and (b) is a graph of calculated x data and z data. It is a flowchart which shows an example of the determination flow of the obstruction based on this invention. It is a graph which shows a brake start distance. FIG. 3 is a scatter diagram in which the horizontal axis represents z data and the vertical axis represents x data. FIG. 4 is a scatter diagram in which the horizontal axis represents z data and the vertical axis represents x data. FIG. 5B is a scatter diagram in which the horizontal axis represents z data and the vertical axis represents x data. FIG. 6B is a scatter diagram in which the horizontal axis represents z data and the vertical axis represents x data. FIG. 7B is a scatter diagram in which the horizontal axis represents z data and the vertical axis represents x data. It is a flowchart which shows an example of the determination flow of the obstruction which concerns on 2nd Embodiment. It is a flowchart which shows an example of the determination flow of the obstruction which concerns on 3rd Embodiment. (A) is a side view showing a detection range when no obstacle is present, and (b) is a graph regarding xz and a graph regarding x-reflection intensity. (A) is a side view showing a detection range when an obstacle is present, and (b) is a graph regarding xz and a graph regarding x-reflection intensity. (A) is a side view which shows the detection range in case a fine particle exists, (b) is a graph regarding xz and a graph regarding x-reflection intensity. It is a perspective view which shows the experiment example of an obstruction detection apparatus. 6 is a graph related to x-reflection intensity measured in the experimental example of FIG. 5, (a) relates to data before correction of reflection intensity, and (b) relates to data after correction of reflection intensity. 6 is a graph regarding xz measured in the experimental example of FIG. 5, (a) relates to data before correction of reflection intensity, and (b) relates to data after correction of reflection intensity. 6 is a graph regarding yz measured in the experimental example of FIG. 5, (a) relates to data before correction of reflection intensity, and (b) relates to data after correction of reflection intensity. 6 is a graph related to y-x measured in the experimental example of FIG. 5, where (a) relates to data before correction of reflection intensity, and (b) relates to data after correction of reflection intensity. It is a flowchart which shows an example of the determination flow of the obstruction which concerns on 4th Embodiment.

  In FIG. 1, an obstacle detection apparatus 1 according to the present invention is mounted on a construction vehicle such as a rolling roller that operates while traveling at a low speed. FIG. 1 shows a case where the obstacle detection device 1 is mounted on a tire roller 10 that performs rolling pressure on an asphalt road surface or the like with a tire 11. The obstacle detection device 1 includes a distance image sensor (3D distance sensor) 2 of a TOF (Time Of Flight) method for measuring a distance from a time difference between projected light and reflected light, and an obstacle based on measurement data of the distance image sensor 2. And a control device 3 for determining the presence or absence of the object G.

"Distance image sensor 2"
The distance image sensor 2 includes a light emitting unit that emits projection light such as infrared rays, and a light receiving unit that receives reflected light when the projection light hits an object. The distance to the object is measured by measuring the time from when the infrared ray is transmitted from the light emitting unit to when the reflected light is received by the light receiving unit. The projection angle from the distance image sensor 2 is, for example, a horizontal angle of 95 °, a vertical angle (symbol θ1 shown in FIG. 1B) is 32 °, and the projection cross section has a horizontally long rectangular shape. The image resolution is, for example, a total of 1024 pixels, 64 pixels in the horizontal direction and 16 pixels in the vertical direction. The distance image sensor 2 is attached to the center of the rear portion of the tire roller 10 in the vehicle width direction so that the projection light is projected obliquely downward in the reverse direction.

  Regarding the detection range of the obstacle G, if the projection range of the projection light is set as the detection range as it is, that is, if the dimension W in the vehicle width direction is set wider than the vehicle width dimension of the tire roller 10, there is no risk of collision. First, it is recognized that there is an obstacle G, and the vehicle stops uselessly. Therefore, it is preferable that the dimension W of the detection range 4 (indicated by hatching in FIG. 1) in the vehicle width direction is set to be substantially the same as the vehicle width dimension of the tire roller 10. Since the distance image sensor 2 can measure the distance to the obstacle G, the obstacle G exists in the detection range 4 set in the vehicle width dimension from the measurement data for each pixel, specifically, y data described later. The control device 3 can determine whether or not to do so. By using the distance image sensor 2 in this way, the dimension W of the detection range 4 can be ensured uniformly over the vehicle longitudinal direction. That is, the detection range 4 can be easily set to a substantially rectangular range having one side as a dimension W in plan view as shown in FIG. The dimension L2 in the vehicle front-rear direction of the detection range 4 is appropriately set according to a commonly used traveling speed, and is set to, for example, about 3 meters in the present embodiment.

  Further, since the projection light of the distance image sensor 2 is projected obliquely downward in the backward direction, the lateral angle θ2 of the projection light when viewed in plan is in a range larger than 95 °. Therefore, with respect to the non-detection ranges 5 and 5 formed between the rear ends of the tire roller 10 and the detection range 4, the distance L3 in the vehicle front-rear direction can be kept small. That is, the non-detected blind spot formed on both sides of the rear portion of the vehicle can be reduced.

"Control device 3"
In FIG. 1, the control device 3 is mounted, for example, near the distance image sensor 2 or around the driver's seat. From the measurement data of each pixel measured by the distance image sensor 2, the control device 3 determines the backward direction component from the rear end of the vehicle as x data, the vehicle width direction component from the center of the vehicle width as y data, and from the ground. Z data is calculated for the height direction component. Then, the control device 3 performs the following three steps when determining the obstacle G.

"Height selection step"
The control device 3 performs a height selection step in which only the measurement data whose z data is larger than the threshold value T1 is selected to determine the presence or absence of the obstacle G. This height selection step is performed mainly for the purpose of preventing the slight unevenness of the ground from being recognized as the obstacle G. The threshold value T1 is set in the range of 5 to 30 cm, preferably in the range of 10 to 20 cm. If the threshold value T1 is set in the range of 5 to 30 cm, it is possible to ignore the generally assumed unevenness of the ground and reduce unnecessary obstacle detection. Further, when the posture of the worker or the like becomes low within the detection range 4, the height can hardly be 30 cm or less, and at this time, it can be considered that there is a possibility of the obstacle G.

"First judgment step"
The control device 3 calculates the inclination of the straight line of the distribution from the x data and the z data of the measurement data selected in the height selection step, and determines the presence or absence of the obstacle G based on this inclination. In the present embodiment, the slope m is calculated by the least square method. Then, the control device 3 determines that there is an obstacle G when the slope m is larger than the threshold value T2, smaller than 0, or division by zero. The slope m by the least square method was obtained by the equation (1).

  In construction vehicles, when working on a slope such as a slope or loading on a trailer, the control device 3 may erroneously determine the surface of the slope as an obstacle G when approaching an uphill from a flat surface. The first determination step is performed for the purpose of preventing the uphill from being recognized as the obstacle G. Generally, the limit climbing angle of a rolling roller and other construction vehicles is about 30 °, and a special one is about 35 °. The slope m of the 30 ° slope according to the least square method is about 0.6, and the slope m of the 35 ° slope is about 0.7. The value of the slope slope m according to the least square method increases as the slope angle increases. Therefore, if the threshold value T2 is set in the range of 0.6 to 0.7, it is possible to avoid useless obstacle detection for an uphill that can be generally assumed. And since the inclination m regarding the person or object standing up and down is generally larger than 0.6-0.7, these can be recognized as the obstacle G. The threshold value T2 can be changed as appropriate.

  Further, regarding the inclination m of the person standing up and down, for example, when the upper body side is close to the vehicle or the work equipment held in the hand is close to the vehicle, the inclination m is negative, that is, smaller than 0. As for the object, when the upper part is closer to the vehicle than the lower part, the inclination m is smaller than zero. Therefore, even when the slope m is smaller than 0, the obstacle G is recognized. When the measurement result becomes completely vertical, in equation (1), the slope m = 0/0, that is, division by zero, so that the obstacle G is recognized also in this case.

  2 to 6 show test results conducted by the inventor, wherein (a) is a side view showing a detection target, and (b) is a graph of calculated x data and z data. The threshold T1 for the height selection step described above is 10 cm, and z data measurement points (shown by black circles) lower than 10 cm are not taken into consideration for the determination of the obstacle G.

  FIG. 2 shows a case where the detection target is a standing worker. FIG. 2A shows a case where 7 pixels out of 16 pixels in the vertical direction detect workers at measurement points p1 to p7 that are 10 cm or more from the ground. The slope m of the measurement points p1 to p7 by the least square method was −7.8485 as shown in FIG. That is, since the slope m is smaller than 0, the control device 3 determines the worker as an obstacle G in the first determination step.

  FIG. 3 shows a case where the detection target is a rectangular box having a height. In this case, the slope m of the measurement points p1 to p7 by the least square method was 6.9676 as shown in FIG. That is, since the slope m is larger than the threshold value T2 in the range of 0.6 to 0.7, the control device 3 determines that the box is an obstacle G in the first determination step.

  FIG. 5 shows a case where the detection target is a slope with a gradient of 18.6 °. The slope m of the measurement points p1 to p16 by all 16 pixels by the least square method was 0.3362 as shown in FIG. That is, since the slope m is smaller than the threshold value T2 in the range of 0.6 to 0.7 and larger than 0, the controller 3 determines that the slope with the slope of 18.6 ° is an obstacle by the first determination step. Do not judge.

  FIG. 6 shows a case where the detection target is a slope with a slope of 4.2 °. The slope m of the measurement points p1 to p16 by all 16 pixels by the least square method was 0.0735 as shown in FIG. In other words, since the slope m is smaller than the threshold value T2 in the range of 0.6 to 0.7 and larger than 0, the control device 3 determines that the slope with a slope of 4.2 ° is an obstacle by the first determination step. Do not judge.

  FIG. 4 shows a case where the detection target is a rectangular box having a low height. FIG. 4A shows a case where 6 pixels detect the top surface of the box at measurement points p1 to p6 and 3 pixels detect the side surface of the box at measurement points p7 to p9. The slope m of these measurement points p1 to p9 by the least square method was 0.2706 as shown in FIG. That is, since the inclination m is smaller than the threshold value T2 in the range of 0.6 to 0.7 and larger than 0, the control device 3 does not determine the box as the obstacle G in the first determination step. End up. This is because, in addition to the measurement points p7 to p9 on the side surface of the box, the measurement points p1 to p6 on the horizontal upper surface of the box are also calculated, and the inclination m is leveled as a whole. For this problem, the control device 3 performs the second determination step.

"Second determination step"
Controller 3, when calculating the coefficient of determination R ² from the x data and z-data of the measurement data having been selected by the height selection step, it is determined that there is no obstacle G in the first determination step, the coefficient of determination R ² Is smaller than the threshold value T3, it is determined that there is an obstacle G. The coefficient of determination R ² is an indicator of the accuracy of the regression analysis, a high correlation higher the number. The coefficient of determination ^{R 2} can be obtained by equation (2).

Usually, slopes often follow almost the same slope, so the correlation between measurement points is high. The slope of Figure 5, the coefficient of determination ^{R 2} measurement points p1~p16 is 0.9677, the slope of FIG. 6, the coefficient of determination ^{R 2} measurement points p1~p16 was 0.8312. The results of the test, the coefficient of determination ^{R 2} of each slope was generally in the range of 0.6 to 1.0. Therefore, the threshold value T3 0.6 to 1.0, preferably is set within the range of 0.7 to 1.0, the control unit 3, the slope of the detection target large coefficient of determination ^{R 2} than the threshold T3 Recognize and do not determine as an obstacle G.

On the other hand, if the box shown in FIG. 4, a 0.4159 determination coefficient ^{R 2} of the measurement points P1 to P9, it was found correlation between the measurement points P1 to P9 is low. Therefore, if you set the threshold T3 in a range of 0.6 to 1.0, the control unit 3, a small box coefficient of determination ^{R 2} than the threshold T3 and obstacle G determines.

  An example of the obstacle determination flow is shown in FIG. The control device 3 calculates x data, y data, and z data from the measurement data of each pixel measured by the distance image sensor 2 (step S1), and the y data for the dimension W of the detection range 4 It is determined whether it is in the range of −W / 2 to W / 2 (step S2). In the case of Yes, it is determined that there is a possibility that the obstacle G exists in the detection range 4, and the process proceeds to step S3. In the case of No, the process returns to step S1.

  In step S3, the control device 3 determines whether or not the z data is larger than the threshold value T1 as a height selection step. If Yes, the process proceeds to step S4, and if No, the process returns to step S1.

The control device 3 calculates the gradient m from the x data and the z data by the least square method, calculates the determination coefficient R ² (step S4), and determines whether the gradient m is greater than the threshold value T2 or greater than 0. It is determined whether or not it is small, or whether it is 0/0, that is, division by zero (step S5). If Yes, it is determined that there is an obstacle G and the process proceeds to step S7. If No, the process proceeds to step S6. In step S6, the control unit 3, a determination coefficient of determination R ² is either smaller or not than the threshold value T3, if Yes, the process proceeds to step S7 it is determined that there is an obstacle G, If No, step Return to S1.

  In step S7, the control device 3 calculates the x data (Min) closest to the vehicle among the x data and the z data, and recognizes this as the distance from the vehicle to the obstacle G. In step S8, the control device 3 determines whether or not the x data (Min) is smaller than a predetermined threshold T4. If yes, the control device 3 transmits a brake signal to a brake means (not shown) and stops the vehicle. (Step S9), if No, return to Step S7. In step S9, an alarm may be issued by sound or light instead of or in combination with the brake.

  In addition, it is preferable to change the timing which outputs a brake signal according to the travel speed of a vehicle. In this case, as shown in FIG. 8, the control device 3 compares the brake start distance S (that is, the threshold value T4) set in advance according to the traveling speed with the x data (Min), and the x data (Min) is When it becomes smaller than the predetermined threshold value T4, a brake signal is output. The brake start distance S is set, for example, to a distance having a slight margin than the limit braking distance T measured for the vehicle. In FIG. 8, the brake start distance S is set to about 0.5 m at a speed of 2 km, about 1 m at a speed of 4 km, about 1.6 m at a speed of 6 km, and about 2.4 m at a speed of 8 km. Note that examples of the vehicle speed sensor that detects the traveling speed of the vehicle include proximity sensors such as a rotary encoder that detects the rotational speed of the tire.

  As described above, if the control device 3 is configured to perform the height selection step of selecting only the measurement data whose z data is larger than the threshold value T1 and determining the presence or absence of the obstacle G, the unevenness of the ground can be ignored. , Wasteful obstacle detection can be reduced.

  The control device 3 calculates the gradient m of the distribution from the x data and the z data of the measurement data selected in the height selection step, and performs a first determination step of determining the presence or absence of the obstacle G based on this gradient m. If it is set as a structure, the possibility of determining an uphill as an obstacle G can be reduced. In particular, if the slope m is calculated by the least square method and the slope m is larger than the threshold value T2, smaller than 0, or divided by zero, it is determined that there is an obstacle G. The G detection accuracy is further improved.

Controller 3, when calculating the coefficient of determination R ² from the x data and z-data of the measurement data having been selected by the height selection step, it is determined that there is no obstacle G in the first determination step, the coefficient of determination R ² If the second determination step of determining that there is an obstacle G when the value is smaller than the threshold value T3, the detection accuracy of the obstacle G is further improved.

  The preferred embodiment of the present invention has been described above. In the embodiment described above, the distance image sensor 2 is attached to the rear part of the vehicle. However, the distance image sensor 2 may be attached to the front part of the vehicle to detect the forward direction of the vehicle.

“Second Embodiment”
In the above-described embodiment, as the first determination step, as shown in FIG. 2B or the like, a scatter diagram in which the horizontal axis is x data and the vertical axis is z data is used. Was used to calculate the slope m. In contrast, in the second embodiment, as a first determination step, the horizontal axis is z data, the vertical axis is a scatter diagram with x data, and the slope a is calculated by the least square method of Equation (3).

  The control device 3 obtains an inclination 1 / a having the inclination a as a denominator. The slope 1 / a of the 30 ° slope according to the least square method of Equation (3) is about 0.6, and the slope 1 / a of the 35 ° slope is about 0.7. Therefore, if the threshold value T4 is set in the range of 0.6 to 0.7, it is possible to avoid useless obstacle detection for an uphill that can be generally assumed. And since inclination 1 / a regarding a person or an object standing up and down is generally larger than 0.6 to 0.7, these can be recognized as an obstacle G. When the upper body side is close to the vehicle or the work equipment or the like held in the hand is close to the vehicle, the inclination 1 / a is negative, that is, smaller than 0. As for the object, when the upper part is closer to the vehicle than the lower part, the inclination 1 / a is smaller than zero. Therefore, even when the slope 1 / a is smaller than 0, the obstacle G is recognized. That is, in the first determination step, the control device 3 determines that there is an obstacle G when the slope 1 / a is larger than the threshold value T4 or smaller than 0.

  9 to 13 are scatter diagrams in which the horizontal and vertical axes of the scatter diagrams in FIGS. 2B to 6B are replaced, respectively, where the horizontal axis is z data and the vertical axis is x data. Yes. As described above, the measurement point (indicated by a black circle) of z data lower than the threshold value T1 is not taken into consideration for the determination of the obstacle G.

  FIG. 9 shows a case where the detection target is a standing worker. The slope a of the measurement points p1 to p7 according to the least square method is −0.0286, and the slope 1 / a is −34.9. That is, since the slope 1 / a is smaller than 0, the control device 3 determines the worker as an obstacle G in the first determination step.

  FIG. 10 shows a case where the detection target is a rectangular box having a height. The slope a of the measurement points p1 to p7 according to the least square method is 0.0831, and the slope 1 / a is 12.0. That is, since the inclination 1 / a is larger than the threshold value T4 in the range of 0.6 to 0.7, the control device 3 determines the box as the obstacle G in the first determination step.

  FIG. 12 shows a case where the detection target is a slope with a gradient of 18.6 °. The slope a of the measurement points p1 to p16 according to the least square method is 2.8784, and the slope 1 / a is 0.34. That is, since the slope 1 / a is smaller than the threshold value T4 in the range of 0.6 to 0.7 and larger than 0, the control device 3 obstructs the slope with the slope of 18.6 ° in the first determination step. Not judged as a thing.

  FIG. 13 shows a case where the detection target is a slope with a slope of 4.2 °. The slope a of the measurement points p1 to p16 according to the least square method is 11.303, and the slope 1 / a is 0.088. That is, since the slope 1 / a is smaller than the threshold value T4 in the range of 0.6 to 0.7 and larger than 0, the control device 3 obstructs the slope with the slope of 4.2 ° in the first determination step. Not judged as a thing.

FIG. 11 shows a case where the detection target is a rectangular box having a low height. In FIG. 4 (b), the slope m by the least square method of the measurement points p1 to p9 is 0.2706, which is smaller than the threshold value T2 in the range of 0.6 to 0.7 and larger than 0. . Thus, in the first determination step controller 3 for not determined the box with the obstacle G, it has required comparison determination of the coefficient of determination R ² in the next second determination step.

On the other hand, according to the second embodiment, in FIG. 11, the inclination a of the measurement points p1 to p7 by the least square method is 1.5372, and the inclination 1 / a is 0.65. If the threshold value T4 is set to 0.6, which is a slope value of 30 °, the slope 1 / a becomes a value larger than the threshold value T4, and the control device 3 sets the box as an obstacle G in the first determination step. It will be judged. Therefore, in this case, a second determination step of comparing the coefficient of determination R ² is not required.

  The inventor conducted a number of tests on a relatively small box, other flat objects, etc. in addition to the example of FIG. As a result, as compared with the first determination step using the slope m calculated by the method of least squares in a scatter diagram in which the horizontal axis is x data and the vertical axis is z data, the horizontal axis The first determination step using the slope 1 / a calculated by the least square method in a scatter diagram with z data and the vertical axis as x data improves the accuracy of distinguishing between a slope and an obstacle G. There was found.

  An example of the obstacle determination flow in the second embodiment is shown in FIG. Steps S1 to S3 and S7 to S9 are the same as those in FIG. In step S11, the control device 3 calculates the slope a from the x data and the z data by the least square method, and in step S12, whether the slope 1 / a is larger than the threshold value T4 or smaller than 0. Make a decision. If Yes, it is determined that there is an obstacle G and the process proceeds to Step S7. If No, the process returns to Step S1.

Also in the second embodiment, it is acceptable to configure the second determination step for comparing the coefficient of determination R ² and the threshold T3. In this case, when No in step S12 in FIG. 14, step S6 similar to that in FIG. 7 may be executed.

“Third Embodiment”
In the second embodiment, as the first determination step, the horizontal axis is z data, the vertical axis is a scatter diagram of x data, the slope a is calculated by the least square method of Equation (3), and this slope a is defined as the denominator. The slope 1 / a to be compared with the threshold value T4. In contrast, the third embodiment is a scatter diagram in which the horizontal axis is z data and the vertical axis is x data, and is the same as the second embodiment until the slope a is calculated by the least square method of Equation (3). There is a difference from the second embodiment in that the slope a is directly compared with the threshold value T5 without fractionation. In the third embodiment, it is determined that there is an obstacle G when the slope a is smaller than the threshold value T5.

  When the slope of the slope is θ, the threshold T5 is obtained by “tan (90−θ)”. θ is 30 ° and the threshold T5 is about 1.73, and θ is 35 ° and the threshold T5 is about 1.43. Therefore, by setting the threshold value T5 in the range of 1.4 to 1.7, it is possible to improve the accuracy of discrimination between the slope and the obstacle G. Hereinafter, the scatter diagrams of FIGS. 9 to 13 are verified for the case where the threshold value T5 is set to 1.7, which is the value of the slope of 30 °.

Since the slope a in FIG. 9 is −0.0286, it is smaller than the threshold value T5. Thereby, the control device 3 determines the worker as an obstacle G.
Since the slope a in FIG. 10 is 0.0831, it is smaller than the threshold value T5. Thereby, the control apparatus 3 determines the box as an obstacle G.
The slope a in FIG. 12 is 2.8784, which is larger than the threshold value T5. Thereby, the control apparatus 3 does not determine the slope of 18.6 degrees as an obstacle.
The slope a in FIG. 13 is 11.303, which is larger than the threshold value T5. As a result, the control device 3 does not determine a slope with a slope of 4.2 ° as an obstacle.
Since the slope a in FIG. 11 is 1.5372, it is smaller than the threshold value T5. Thereby, the control apparatus 3 determines the box as an obstacle G. Therefore, similarly to the second embodiment, the second determination step of comparing the coefficient of determination R ² is not required. As a result of verifying other objects, it has been found that the accuracy of distinguishing between the slope and the obstacle G is improved also in the third embodiment.

  An example of the obstacle determination flow in the third embodiment is shown in FIG. Steps S1 to S3, S11, and S7 to S9 are the same as those in FIG. The control device 3 determines whether or not the inclination a is smaller than the threshold value T5 in step S13. If Yes, it is determined that there is an obstacle G and the process proceeds to Step S7. If No, the process returns to Step S1.

Also in the third embodiment, it is acceptable to configure the second determination step for comparing the coefficient of determination R ² and the threshold T3. In this case, when No in step S13 in FIG. 15, step S6 similar to FIG. 7 may be executed.

“Fourth Embodiment”
The control device 3 performs a “height selection step” in which only the measurement data whose z data is larger than the threshold value T1 is selected in each step S3 of FIG. 7, FIG. 14, and FIG. Instead of this “height selection step”, the control device 3 of the fourth embodiment selects only measurement data whose reflection intensity of reflected light is greater than the threshold value T6 and determines whether there is an obstacle “reflection intensity selection” Step "is performed. That is, the control device 3 determines the presence or absence of the obstacle G based on the measurement data regarding the coordinates measured by the distance image sensor 2 and the reflection intensity of the reflected light.

  Examples of objects that can be detected by the distance image sensor 2 include water vapor and dust in addition to people and objects that obstruct travel. When rolling down the asphalt road surface with a rolling roller, in order to prevent the asphalt mixture from adhering to the rolling wheel of a tire, etc., the asphalt adhesion agent or water is sprayed on the rolling wheel while rolling. There are many cases. These liquid agents may generate water vapor by touching hot asphalt road surfaces and tire surfaces. In addition, when compacting ground soil, dust may rise from the ground. It is not preferable from the viewpoint of work efficiency to determine these water vapor and dust as obstacles.

  In response to this problem, the present inventor has focused on the reflection intensity of the reflected light and decided to improve the determination accuracy of the obstacle G by adding a correction function of the reflection intensity. The reflection intensity of the reflected light varies depending on the distance, shape, material, color tone, and the like of the object, and the reflection intensity for water vapor and dust is lower than that of a person or a solid object. The present inventor analyzed the following as focusing on the reflected light from the ground.

  As shown in FIG. 16A, when there is no obstacle in the measurement range of the distance image sensor 2, all the coordinate data measured are related to the ground. In this case, as shown in FIG. 16B, the graph S1 indicating the xz relationship shows that z data, that is, the height direction component is 0 over all x data, and there is no obstacle. . On the other hand, the reflection intensity of the reflected light tends to decrease as the distance between the measurement point and the distance image sensor 2 increases. Therefore, in the graph P1 indicating the relationship of x-reflection intensity, the reflection intensity of the reflected light from the ground decreases as the x data increases, that is, as the distance from the vehicle increases.

  As shown in FIG. 17A, when the obstacle G exists within the measurement range of the distance image sensor 2, as shown in FIG. 17B, the graph S2 showing the relationship of xz is an obstacle. The z data changes according to the measurement point of G. Thereby, it turns out that the obstruction G exists. The graph P2 indicating the relationship between the x-reflection intensity depends on the color tone of the obstacle G, but the reflection intensity of the reflected light from the obstacle G is generally higher than the reflection intensity from the original ground.

  As shown in FIG. 18A, when a fine particle F such as water vapor or dust floats in the measurement range of the distance image sensor 2, as shown in FIG. It is represented as a plot S3 scattered almost randomly. Depending on the distribution state of the plot S3, water vapor, dust, etc. may be determined as an obstacle. On the other hand, regarding the relationship between the x-reflection intensity, the reflection intensity of reflected light from water vapor or dust is low, so the graph P3 is comparable to the graph P1 in FIG. 16B or lower than the graph P1. It turns out that it becomes a tendency. Therefore, when the reflection intensity of the graph P1 in FIG. 16B, that is, the reflection intensity of the reflected light from the ground is set as the threshold T6 and the reflection intensity of the reflected light from the object is equal to or less than the threshold T6, the control device 3 If it is assumed that the object is a floating fine particle F and is not determined as an obstacle G, the determination accuracy of the obstacle G can be improved.

  As the threshold value T6, a fixed value pattern consisting of a constant value obtained in advance by simulation or the like, and a reflection intensity of reflected light from the ground obtained during actual construction operation are calculated in real time, and a variation obtained based on the calculation result Value pattern. In the case of the latter variation value pattern, the threshold value T6 is determined by changing in real time based on the material and color tone of the road surface that is actually rolled, so that the accuracy of the threshold value T6 is increased, and water vapor, dust, and the like are removed from the obstacle G. Since the accuracy of the cancel function to be excluded can be further increased, the detection accuracy of the obstacle G can be increased as a result. Further, the operator may be configured to manually switch between the fixed value pattern and the variation value pattern, or the fixed value pattern and the variation value pattern may be automatically determined by the control device 3 and switched.

  As shown in FIG. 19, the present inventor measured the water vapor 6, the desk 7, the person 8, and the ground 9, which are ejected from the humidifier 21, with the distance image sensor 2. 20 to 23 show the measurement data. FIG. 20 is a graph relating to the x-reflection intensity. FIG. 20A shows a state in which the reflection intensity is not corrected at all, and a portion surrounded by a dotted line shows measurement data of the ground 9. FIG. 20B is a graph showing only the measurement data with the reflection intensity higher than the threshold T6, with the measurement data of the ground 9 as the threshold T6. The threshold T6 is preferably a value obtained by adding the standard deviation σ to a value obtained by the least square method of measurement data, and the standard deviation σ is preferably set to 1.5σ to 3σ.

  FIGS. 21, 22, and 23 are graphs relating to xz, yz, and yz, respectively. In these figures, each (a) is a graph when the reflection intensity is not corrected, and each (b) is a graph when the reflection intensity is corrected by the threshold T6. In FIG. 21A, FIG. 22A, and FIG. 23A before correction, the measurement data of the water vapor 6, the desk 7, the person 8, and the ground 9 are displayed, but the threshold T6 is used. In FIG. 21 (b), FIG. 22 (b), and FIG. 23 (b) after correction, only the measurement data of the desk 7 and the person 8 are displayed, and the measurement data of the water vapor 6 and the ground 9 are canceled. I understand. Of course, since the obstacle to be detected in the construction vehicle is an object having a certain height such as the desk 7 or the person 8, there is no problem even if the measurement data of the ground 9 is canceled by the correction by the threshold T6. Absent.

  An example of the obstacle determination flow of the fourth embodiment is shown in FIG. Steps S1 and S2 are the same as those shown in FIGS. In step S31, the control device 3 determines whether or not the reflection intensity of the measurement data of each pixel is higher than the threshold value T6. If yes, the control device 3 proceeds to step S4 in FIG. 7 or step S11 in FIG. If No, go back to Step S1.

  As described above, if the configuration is such that the presence or absence of the obstacle G is determined based on the measurement data relating to the coordinates measured by the distance image sensor 2 and the reflection intensity of the reflected light, such as water vapor or dust floating in the air. Misidentification of the fine substance F as the obstacle G is reduced. Thereby, useless obstacle detection can be suppressed.

  Further, if the control device 3 is configured not to determine the obstacle G when the reflection intensity is equal to or lower than the threshold T6, the control device 3 erroneously determines the fine particle F as the obstacle G by setting the threshold T6 to an appropriate value. This can be reduced with a simple configuration.

The reflection intensity of the reflected light from the fine particles F is almost the same as or less than the reflection intensity of the reflected light from the ground. Therefore, the threshold T6 is set to a value based on the reflection intensity of the reflected light from the ground. The erroneous determination of the object F as the obstacle G can be further reduced.
Since it is possible to prevent erroneous determination of unevenness on the ground by correcting the reflection intensity, a “height selection step” is not required.

DESCRIPTION OF SYMBOLS 1 Obstacle detection apparatus 2 Distance image sensor 3 Control apparatus 4 Detection range 5 Non-detection range 10 Tire roller (construction vehicle)

Claims (15)

An obstacle detection device mounted on a construction vehicle,
A TOF-type distance image sensor that measures the distance from the time difference between the projected light and the reflected light;
A control device for determining the presence or absence of an obstacle based on the measurement data of the distance image sensor;
With
Regarding the measurement data, when the vehicle traveling direction component from the vehicle end is x data, the vehicle width direction component is y data, and the height direction component from the ground is z data,
The obstacle detection device for a construction vehicle, wherein the control device performs a height selection step of selecting only measurement data whose z data is larger than a threshold value T1 and determining the presence or absence of an obstacle. An obstacle detection device mounted on a construction vehicle,
A TOF-type distance image sensor that measures the distance from the time difference between the projected light and the reflected light;
A control device for determining the presence or absence of an obstacle based on the measurement data of the distance image sensor and the reflection intensity of the reflected light;
With
The obstacle detection device for a construction vehicle, wherein the control device performs a reflection intensity selection step of selecting only measurement data in which the reflection intensity of the reflected light is larger than a threshold value T6 and determining the presence or absence of an obstacle.   The controller calculates a slope of the distribution from the x data and the z data of the measurement data selected in the height selection step, and performs a first determination step of determining the presence or absence of an obstacle based on the inclination. The obstacle detection device for a construction vehicle according to claim 1, wherein the obstacle detection device is a vehicle.   The controller calculates a slope of the distribution from the x data and the z data of the measurement data selected in the reflection intensity selection step, and performs a first determination step of determining the presence or absence of an obstacle based on the inclination. The obstacle detection device for a construction vehicle according to claim 2, characterized in that:   The construction vehicle according to claim 3 or 4, wherein the inclination is an inclination m calculated by a least square method in a scatter diagram of the x data on the horizontal axis and the z data on the vertical axis. Obstacle detection device.   6. The obstacle detection device for a construction vehicle according to claim 5, wherein it is determined that there is an obstacle when the slope m is larger than a threshold T2, smaller than 0, or division by zero.   The construction vehicle according to claim 3 or 4, wherein the inclination is an inclination a calculated by a least square method in a scatter diagram of the z data on the horizontal axis and the x data on the vertical axis. Obstacle detection device.   The obstacle detection device for a construction vehicle according to claim 7, wherein it is determined that there is an obstacle when the slope 1 / a obtained from the slope a is larger than a threshold value T4 or smaller than 0.   The obstacle detection device for a construction vehicle according to claim 7, wherein it is determined that there is an obstacle when the inclination a is smaller than a threshold value T5.   The control device calculates a determination coefficient from the x data and the z data of the measurement data selected in the height selection step, and when it is determined in the first determination step that there is no obstacle, the determination coefficient is a threshold T3. The obstacle detection device for a construction vehicle according to claim 3, wherein a second determination step for determining that there is an obstacle is performed when the obstacle is smaller than the threshold.   The said threshold value T1 is the range of 5-30 cm, The obstacle detection apparatus of the construction vehicle of Claim 1 or Claim 3 characterized by the above-mentioned.   The said threshold value T2 is the range of 0.6-0.7, The obstacle detection apparatus of the construction vehicle of Claim 6 characterized by the above-mentioned.   The said threshold value T4 is the range of 0.6-0.7, The obstacle detection apparatus of the construction vehicle of Claim 8 characterized by the above-mentioned.   The said threshold value T3 is the range of 0.6-1.0, The obstacle detection apparatus of the construction vehicle of Claim 10 characterized by the above-mentioned.   The obstacle detection device for a construction vehicle according to claim 9, wherein the threshold value T5 is in a range of 1.4 to 1.7.

Images