JP2683899B2 - Unmanned carrier - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Outline] The present invention relates to a line tracking system, in particular, a line tracking system in which a vehicle body is driven along a marking line while an image of a marking line on a road surface is picked up by a TV camera attached to the vehicle body. For the purpose of providing an unmanned transportation device capable of operating an unmanned transportation vehicle with a simple construction process, a means for forming a linear mark on the floor surface and imaging the linear mark on a moving vehicle A processing unit that processes image data of the captured linear marks, and a mechanism unit that drives and steers the moving vehicle based on the processing result in the calculating unit. It is configured to operate automatically. BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a line tracking system, and more particularly to a line tracking system in which a vehicle body is operated along a marking line while imaging a marking line on a road surface with a TV camera attached to the vehicle body. is there. [Prior Art] In a factory or a place for carrying various kinds of luggage, an unmanned carrier may be operated to improve or improve the work efficiency of carrying luggage by human driving. However, a conventional example of such an unmanned transportation system is shown in FIG. 13, for example. This is a rail 2 in a groove formed on a floor 1 such as a road surface.
On the other hand, a current collector 4 is attached to the transport vehicle 3 to receive electric power from the current collector 4 in contact with the rail 2 and the transport vehicle 3 travels unmanned while being guided by the rail 2. Is. [Problems to be Solved by the Invention] However, in such a conventional unmanned transport system, when it is desired to newly install or change the traveling route of the transport vehicle 3, the floor surface 1 is There is a problem in that construction or renovation procedures become large-scale work because it requires construction work. The present invention has been made in view of such conventional problems, and an object of the present invention is to provide an unmanned transportation device capable of driving an unmanned transportation vehicle by an extremely simple construction process. [Means for Solving the Problems] FIGS. 1 to 3 show the principle of the present invention. FIG. 1 is a view showing various linear marks 5 provided on the floor surface 1 of a road or a premises. (A) is a white linear mark 5a provided on the floor surface 1,
(B) shows the linear marks 5b colored in yellow or the like. Also, FIG. 1 (c) shows an example in which linear marks 5a and 5b are paired and applied to the floor surface 1, and FIG. 1 (d) is a line drawn on the floor surface 1 at cut intervals of a predetermined dimension or less. Showing a circular mark 5c. FIG. 2 shows a mobile vehicle used in the unmanned transportation device of the present invention. The moving vehicle 6 has an image pickup means 7 such as a TV camera mounted on the front end of the vehicle body, and the image pickup means 7 picks up images of the linear marks 5a, 5b, 5c formed on the floor surface. . FIG. 3 is a view showing an image pickup field of the image pickup means 7. As is clear from this figure, the image pickup means is oriented so as to catch the linear mark 5 in the substantially central portion of its visual field 8. Further, in the moving vehicle 6, as a calculation unit 9 for processing the image data of the linear mark imaged as shown in FIG.
A means (corresponding to the filter 13) for extracting data corresponding to the linear mark from the imaged image data, and a dividing means (n division profile) for dividing the image data into a plurality of sections by intersecting the traveling direction of the moving vehicle. (Corresponding to the processing unit 14),
A midpoint extraction unit 15 for extracting the midpoint of the data corresponding to the linear mark from the image data divided into each section, and a line connecting the midpoints between the sections extracted by the midpoint extraction unit 15 Determining means 16 for recognizing the shape of the travel route based on the determination result of the direction component of the vector, and means for controlling the driving and steering of the moving vehicle 6 based on the recognition result by the determining means 16. (Corresponding to the mechanical unit 10 shown in FIG. 2) is provided. [Operation] When the moving vehicle 6 is installed on the traveling track and the operation is started, the image capturing unit 7 captures an image of the linear mark 5. The image data of the linear mark 5 obtained by this imaging is A / D converted and transferred to the calculation unit 9. In the calculation unit 9, the dividing means intersects the image data with the traveling direction of the moving vehicle 6 to divide the image data into n, and the midpoint extracting means 15 extracts the midpoint of the line mark for each of the divided sections. Then, the judgment means 16 recognizes the shape of the traveling route based on the judgment result of this direction component by connecting these midpoints to the vector of the linear mark, and transmits the recognition result to the mechanical section 10. And mechanical part
Reference numeral 10 drives and steers the moving vehicle 6 based on the processing data from the calculation unit 9. [Embodiment] FIGS. 4 to 12 are views showing an embodiment of the present invention. FIG. 4 is a diagram showing a specific example of the image pickup unit 7 and the arithmetic unit 9 that performs arithmetic processing based on image data picked up by the image pickup unit 7. The output terminal of the image pickup unit 7 has an A / D
A converter 11 is connected, and the A / D converted image data is transmitted to the calculation unit 9 here. The calculation unit 9 includes a frame memory 12 for temporarily storing the digitized image data, a filter 13 for extracting data corresponding to the linear mark 5 from the output data from the frame memory 12, and the output data for the mobile vehicle 6. N division profile processing unit 14 that divides the image component of each division divided into n perpendicular to the traveling direction, a midpoint extraction unit 15 that extracts the midpoint of the data corresponding to the linear mark 5, and the midpoint The determination unit 16 determines a direction (straight or curved) in which the linear mark 5 extends by connecting the midpoints between the sections extracted by the extraction unit 15. For the filter 13, for example, a median filter is used. FIG. 5 is a diagram showing a specific example of the n-division profile processing unit 14, which is a main element for implementing the present invention. This n
The division profile processing unit 14 includes an X counter 17 for counting coordinates in the X direction in response to the input of the clock signal CL,
A Y counter 18 that receives the output of the X counter 17 and counts coordinates in the Y direction for each predetermined X count value, and a Y counter
An n division detection unit 19 for detecting whether or not the Y count value output by 18 is Y _max / n, a storage area is divided into n sections, and data write input and read output are performed. Switching between the memory 24 and a selector 23 for selecting addressing for the memory 24 by turning the switching signal on and off between the address information input terminal A sent from the CPU and the count value input terminal B from the X counter 17. Inverter 20 for inverting the signal and this inverter 20
AND gate 21 which opens and closes the gate of the write enable WE for the memory 24 by the inverted switching signal from the AND gate, and the AND of the output of the AND gate 21 and the output of the n division detection unit 19 is input to the memory 24. It consists of a gate 22. The operation of the unmanned transportation device having such a configuration will be described. First, when the moving vehicle 6 travels, the image signal obtained by the image pickup by the image pickup unit 7 is converted into a digital signal by the A / D converter 11 and stored in the frame memory 12 as image data. The image data read from the frame memory 12 is noise-removed by the filter 13 and sent to the n-division profile processing unit 14. In the n-divided profile processing unit 14, the input signal of the selector 23 is set to B and the AND gate 21 is turned on by setting the switching signal C to OFF (that is, "0"). On the other hand, the X counter is synchronized with the clock signal CL.
The image data is sent to the data input terminal of the memory 24 in accordance with the designation of the X coordinate address of the memory 24 in sequence, and writing is performed under the ON state of the write enable WE. When the count value of the X counter 17 matches the maximum address value of the X coordinate in the memory 24, the Y counter 18 counts up by 1, and the memory 24
After the line is changed, the image data is again written to the address designated by the X counter 17. By repeating this, writing to the memory 24 is sequentially performed, but in order to divide the image data into n, the count value of the Y counter 18 is checked by the n division detecting unit 19 to detect Y _max / n. For example, assuming that the maximum value of the Y coordinate address of the memory 24 is 512, and if the Y division direction is divided into four, the n division detection unit 19 has Y count values of 0 to 127,128 to 255,256 to 383,38.
As shown in FIG. 6, by dividing the write enable WE at each boundary address into four sections of 4 to 511, the memory 24 is divided into four sections by cross sections a, b, c, d, e. Write divided data. FIG. 6 shows a data storage state in the memory 24 of the n-division profile processing unit 14 and a processing state in the midpoint extraction unit 15 when a linear mark extending linearly is imaged. In this example, the state where the straight line portion of the linear mark 5 is grasped is shown. When performing the midpoint extraction, first, the switching signal C is turned on (that is, "1") to set the input terminal of the selector 23 to A and the AND gate 21 is turned off. As a result, the write enable WE is cut, the address information from the CPU is input to the memory 24, the memory 24 is read, and the read image data is output from the data output terminal. Then, this output image data is sent to the midpoint extraction unit 15. In the midpoint extraction unit 15, a cross section a between the four divided sections
Each linear mark component 26 of e is detected, and the midpoint of the linear mark component is obtained. The midpoint information thus obtained is output from the midpoint extraction unit 15 to the determination unit 16
Output to The determination unit 16 sequentially connects the midpoints obtained in each cross section and the midpoints to obtain the vector of the linear mark 5. Then, when this vector is superimposed on the input image, it becomes as shown in FIG. That is, by this determination operation, the directional components V _a , V _b , V _c , and V _d are obtained from the imaging field of view closer to the vehicle body of the moving vehicle 6. As a result, the determination unit 16 recognizes that the linear mark 5 is linear and issues an instruction to the mechanical unit 10, so that the moving vehicle 6 goes straight. When the linear mark 5 remains straight, as the moving vehicle 6 goes straight, a mark image which is continuously straight within the field of view shown in FIGS. 6 and 7 is obtained. The direction component obtained by extraction is a straight line. Arithmetic unit 9
Then, as the image data of each of the four sections is updated as the moving vehicle 6 moves forward, the four-point profile processing is performed every predetermined time to perform midpoint extraction, and the direction component of the linear mark 5 is determined. A determination is made to control the automatic operation of the moving vehicle 6. The case where the linear mark 5 extending straight is a broken line is processed by the following procedure. That is, as shown in FIG. 8, when a cutting line is included in the visual field 8, the image data including the missing portion 27 is shown in FIGS. Move as in b). And an image of the linear mark 5 is applied to FIG. 8 (a) in all cross sections a to e, the middle point extraction is done for each section a to e, the direction is vector representation components V _a, V _b , V _c , V _d
Is obtained. However, in FIG. 8 (b), the linear mark component 26 cannot be detected in the cross section c, and therefore the midpoint extraction is also impossible. In such a case, the midpoint extraction unit uses the vector address of the previous frame as a temporary candidate for the midpoint on the cross section c.
15 and outputs the direction components V _{b and} V _c as a vector.
As a result, a continuous direction component is obtained, and the moving vehicle 6
Is automatically controlled. Next, the identification of the linear mark 5 on the curve will be described. FIGS. 9 (a) and 9 (b) are diagrams showing the four-division profile processing for the image data when the moving vehicle 6 is approaching a curve and the vector display of the direction component by the midpoint extraction. When the moving vehicle 6 approaches a curve from a straight line path, a curve appears in the image of the linear mark 5 in the section that is the farthest position from the moving vehicle 6 in the section of the field of view 8 that is divided into four, and the direction is displayed as a vector. The component V _d extends at an angle with respect to the preceding direction component V _c .
This judging unit 16 grasps the angle from the vector address.
While the curve gradually becomes clear as the moving vehicle 6 advances, the direction component of the vector display obtained by extracting the midpoint of the linear mark 5 that draws this curve points in the direction of the curve, and accordingly the moving vehicle automatically moves. Be driven. Next, the detection of the T-junction will be described. FIG. 10 (a),
(B) and (c) show image changes when the moving vehicle 6 reaches a T-shaped road. FIG. 10 (a) shows a state in which the T-shaped portion is first caught in the visual field. At this time, since the linear marks 5 are applied to all the cross sections a to e, the midpoint is extracted for each of the cross sections a, b, c, d, and the direction component is displayed as a vector. When the T-shaped portion is completely within the field of view as shown in FIG. 10 (b) due to the forward movement of the moving vehicle 6, the cross-section e is not covered with the linear mark 5. In such a case, for the cross-section e, the vector address one frame before is output as a temporary candidate, and the direction component connecting the midpoints is represented by a vector. Further, when the image of the T-shaped portion gradually moves downward as shown in FIG. 10 (c) due to the movement of the moving vehicle 2, the image of the linear mark 5 between several frames cannot be obtained in the cross section e. In such a case, the output of the vector address between the section d and the section e is terminated. Based on this, the mechanical unit 10 controls the stop of the moving vehicle 6 or the right turn or the left turn at the T-shaped road. Further, the identification of the linear mark 5 in the angled T-shape will be examined (Fig. 11). When the moving vehicle 6 approaches an angled T-shaped road (in this case, when viewed in plan, the upper right corner), first
As shown in FIG. 11 (a), a linear mark (especially 5a in this case) corresponding to an oblique T-shaped portion first appears in the cross section e. When performing midpoint extracted in this order section e, midpoint appear in two places C ₁ and C _2, two vectors V _d connecting the middle point in the cross-section d midpoint C _1, C _2, V _p Address occurs between section d and section e. After that, an oblique T-shaped road gradually appears as the moving vehicle 6 progresses, two vector addresses appear in the section d as shown in FIG. 11 (b), and further the section c as shown in FIG. 11 (c). Two vector addresses appear at the same time, and so on. Then, when the image of the T-shaped portion gradually moves downward, the image of the linear mark 5 (or 5a) cannot be obtained over several frames in the cross section e, and the output of the vector address between the cross section d and the cross section e is output. Aborted. Based on this, the mechanical unit 10 controls the stopping of the moving vehicle 6 or the right turn or the left turn at the T-shaped road. Further, the identification of the linear mark 5 at the crossroads will be examined (Fig. 12). First, as shown in FIG. 12 (a), each cross section a
In e to e, a single vector address is output and a linear vector direction component is created. Next, as shown in FIG. 12 (b), when the cross portion is applied to the cross section d, the midpoints appear at two points (closer to the center and closer to the left end) in the cross section d, and the vector address output produces vertical lines. Triangular vectors V _p and V _q protruding leftward with respect to the vectors V _{a to} V _d corresponding to the circular mark 5 are generated. Then, as the moving vehicle 6 continues to move, the place of appearance of the midpoint in the cross section d changes (gradually toward the center and toward the right end), and the vector address output produces V _a corresponding to the linear mark 5 in the vertical direction. Triangular vectors V _s and V _t protruding to the right with respect to V _d are generated. The mechanical unit 10 recognizes this change in the vector display and performs an optimum operation on the crossroads. [Effects of the Invention] As described above, according to the present invention, the linear mark provided on the floor surface is imaged, and the midpoint of the linear mark is extracted from the image data obtained by this imaging. Since the vector of linear marks is connected by connecting the midpoints and the direction component of the linear mark is determined from this, the moving vehicle can operate along the route based on the calculation processing. Not only straight or curved marks but also sharp turns, angled T-shaped roads, V-shaped roads, etc. can be easily judged, so a simple construction process, inexpensive and highly reliable unmanned It becomes possible to provide a carrier.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a diagram showing various linear marks used for route display in the present invention, FIG. 2 is a diagram showing the principle of an unmanned transporter of the present invention, and FIG. 2 is a diagram showing a visual field of an image picked up by the image pickup unit in FIG. 2, FIG. 4 is a diagram showing an embodiment of a circuit configuration from the image pickup unit to the arithmetic unit in the present invention, and FIG. 5 is used in the embodiment. FIG. 6 is a diagram showing a configuration example of a division profile processing unit, FIG. 6 is a diagram showing a division state of image data in the n division profile processing unit, and a data detection state of a linear mark in each divided cross section;
FIG. 8 is a diagram in which a vector obtained by midpoint extraction and an input image are overlapped with each other, and FIGS. 8A and 8B are diagrams showing a method of identifying a linear mark when a cutting line is mixed, 9 (a) and 9 (b) are diagrams showing a method for identifying curved linear marks, and FIGS. 10 (a), (b), and (c) are diagrams showing a method for identifying T-shaped roads, and FIG. Figures (a), (b) and (c) are diagrams showing a method of identifying a skewed T-shaped road, and Figures 12 (a), (b) and (c).
Is a diagram showing a method of identifying a crossroads, and FIG. 13 is a diagram showing an example of a conventional unmanned transportation device. 1 ... Floor surface, 5 ... Linear mark 6 ... Movable vehicle, 7 ... Imaging unit 8 ... Field of view, 9 ... Calculation unit 10 ... Mechanical unit, 11 ... A / D converter 12 ... Frame memory 13 ... Filter 14 ... n division profile processing unit 15 ... Midpoint extraction unit, 16 ... Judgment unit

Claims (1)

(57) [Claims] An unmanned transportation device that automatically tracks a linear mark on a floor surface and automatically drives a moving vehicle. The unit captures the linear mark and corresponds to the linear mark from the captured image data. Means for extracting data, a dividing means for dividing the image data into a plurality of sections by intersecting the traveling direction of the moving vehicle, and an image obtained by dividing the middle point of the data corresponding to the linear mark into each section. A midpoint extraction means for extracting from the data and a midpoint between the sections extracted by the midpoint extraction means are connected to form a vector of linear marks, and the shape of the travel route is determined based on the determination result of the direction component of the vector. An unmanned transporting apparatus, comprising: a determination unit that recognizes the position of the vehicle, and a unit that controls driving and steering of the moving vehicle based on a recognition result of the determination unit.