JPH01197808A - Guidance system for unmanned vehicle - Google Patents

Abstract

PURPOSE:To make an unmanned car intelligent by analyzing the image information in which a mark has been brought to image pickup, inputting the analyzed image information and an in-building coordinate of the mark in a storage means and calculating the present position and azimuth of the unmanned vehicle, and bringing the unmanned vehicle to running control. CONSTITUTION:A mark 5 is detected by a camera 3, and the center position coordinate of each mark 5 and the contents of a code are read by a vision controller 10. In an arithmetic part 12, the present position and advance direction in a plant coordinate system of an unmanned vehicle 1 are recognized. Based on information which has been recognized by the arithmetic part 12, an instruction of running data is transmitted to a servo-driver 18 by a running control part 14. The control part 14 guides running of the unmanned vehicle by a destination instructed from a communication controller 17, map data 15 in which a running path in a plant has been stored and the information from the arithmetic part 12. In such a way, the unmanned vehicle can be guided without necessitating a complicated image processing.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a guidance device for an unmanned vehicle, and more specifically,
The present invention relates to a device that guides an unmanned vehicle to travel along a desired course based on image information.

[Background of the invention]

The guidance method of the above-mentioned unmanned vehicle (hereinafter simply referred to as unmanned vehicle) is to lay a guidance wire such as an electromagnetic induction wire or optical tape on the floor during the course of the vehicle, and to detect the guidance by detecting the guidance installed on the unmanned vehicle. Many fixed-route guidance systems for unmanned vehicles, which detect the position of unmanned vehicles and control the steering of the unmanned vehicles so that they follow the guide line, are currently in practical use.

However, the fixed route guidance method has a disadvantage in that there is no flexibility in changing the driving course because the driving course is permanently set. Furthermore, the method using electromagnetic induction wires has the problem of requiring a great deal of time and effort to bury the guide wires, and the method using optical tape has the problem that the tape may be damaged due to foot traffic, etc. There is a problem in that guidance errors occur due to dirt or damage.

Therefore, in order to solve the above problems, several new guidance methods other than the fixed route guidance method are currently being considered. One of the new guidance methods is one that guides unmanned vehicles based on image information. Rapid improvements in the performance of image processing technology and microcomputers in recent years, as well as significant reductions in the prices of components such as microcomputers, memory, and A-D/D-A converters, have led to the use of small and inexpensive images in unmanned vehicles. It has become possible to incorporate a processing system and is attracting attention as a next-generation guidance method.

A method for guiding an unmanned vehicle based on the above image information (hereinafter,
For example, as an image guidance method (simply referred to as an image guidance method),
The technology described in Publication No. 144106, that is, an imaging device is placed directly above and along the driving path, the imaging device performs visual recognition of the unmanned vehicle, and detects the amount of driving state of the unmanned vehicle. 2. Description of the Related Art There is a known technique for guiding an unmanned vehicle by transmitting the detection result to the steering and traveling device of the unmanned vehicle via a wireless communication device.

However, in the conventional technology described above, the degree of freedom of the unmanned vehicle is limited to some extent. In other words, even though the price has become cheaper, it is still economically difficult to install an expensive imaging device (including its attached computing device and communication device) in the interior of a building where unmanned vehicles are installed. Naturally, they must be placed along the travel path as described above. Therefore, when changing the layout of a driving route or a building, a problem arises in that the imaging device has to be reinstalled on the new driving path. Furthermore, the basic problem with the above conventional example is that the unmanned vehicle does not know its own position by seeing it with its own eyes, but rather with the eyes of others on the ground. The purpose is to know one's own position by being taught.

In the future, more intelligent driverless cars will inevitably be required to see and make decisions with their own eyes. In other words, rather than simply receiving commands such as ``left,'' ``go straight,'' and ``stop'' from the ground side, the robot determines where its current position is in the entire building or on the planned travel route. By guiding itself by comparing its destination with its current position, it will become more intelligent, and each unmanned vehicle will have its own uniqueness.

Therefore, by equipping each unmanned vehicle with an imaging device, the unmanned vehicle can
A technology that allows humans to "see" is being developed as a next-level guidance method. For example, based on the position information of the unmanned vehicle in the storage means in advance.

Data regarding the arrangement and shape of components is input as a predicted pattern, and the actual image sent from an imaging device on a moving unmanned vehicle is compared with the predicted pattern to reduce the difference. Technology for guiding unmanned vehicles is known.

However, in the image guidance method described above, the storage capacity for storing the predicted pattern is large, the comparison and processing of the predicted pattern with the actual image takes time, and it is difficult to change the layout of the composition. It has been pointed out that there is a problem in that it is necessary to re-input the above predicted pattern every time a driverless vehicle changes course.

Now, in the flow of the prior art as described above, the purpose of the present application is to promote the intelligentization of unmanned vehicles by proposing an image guidance device with a higher degree of freedom.

[Means to solve the problem]

This invention places distinguishable marks at predetermined positions on the ceiling of a building into which unmanned vehicles are introduced, and the unmanned vehicles include (i) means for capturing an image of the ceiling; and (11) captured images. means for analyzing the information; and (iii
) means for storing the intra-building coordinates of each mark; (iv) means for calculating the current position and orientation of the unmanned vehicle by inputting the analyzed image information and the intra-building coordinates of the marks in the storage means; (v) means for controlling the running of the unmanned vehicle based on the output of the calculation means.

〔Example〕

FIG. 1 schematically shows an outline of an unmanned vehicle (1) operating in a building (2) using the image guidance system according to the present invention. The unmanned vehicle (1) according to the present invention is capable of running, steering, etc. using known techniques, and is equipped with a CCD camera (3) as an imaging means on the machine base. Furthermore, distinguishable marks (5) are arranged and pasted at predetermined positions on the ceiling surface (4).

The unmanned vehicle 1) takes an image of the ceiling (4) with the CCD camera (3) (hereinafter simply referred to as the camera), and marks the mark (5).
), it recognizes its current position and direction of travel.

According to the schematic diagram in Figure 2, the camera (3) has a lens (
6) and a screen (7), the screen (
7) Many CCDs! Consists of child columns. The center of the lens (6) and the center of the screen (7) are at the same position in the vertical direction, and a coordinate system (X, y
) is set. The X-axis coincides with the vertical center line of the unmanned vehicle (1), and there is a distance (L) between the y-axis and the horizontal center line (S) of the unmanned vehicle (1). Furthermore, a factory coordinate system (X, Y) within the ceiling plane is set for each mark (5). It is assumed that there is a distance (D) between the lens (6) and the screen (7), and a distance (H) between the lens (6) and the ceiling surface (4). Next, we will explain the concept of measuring the current position and traveling direction of an unmanned vehicle using the above configuration. The mark (5) and the image processing of the mark will be described later.

A>2 Concepts for recognizing the current position and direction of movement of an unmanned vehicle using mark recognition.

The basic idea is that the coordinates (X,
y), first, the camera coordinates (Xo) are obtained by directly translating the coordinates (x, y) to the ceiling surface.

yo) and further converts the camera coordinates (Xo.

y') and the factory coordinates (X, Y) on the ceiling surface, the current position and direction of travel are recognized. below,
I will explain this in more detail, but first I will list the prerequisites for this idea. (i) Camera (3) is mounted on the ceiling (
4) is oriented vertically. (ii) Mark (5)
Create many types of screens and install them at intervals of about half the field of view of the camera (3) so that the screen (7) of the camera (3) is always visible.
It is assumed that at least two marks are imaged within the area.

(iii) If three or more marks are imaged, any two marks are adopted and the others are ignored. (iv) It is assumed that the distance between the camera (3) and the ceiling surface (4) is known and constant. If there is variation, it is assumed that the height of each mark from the floor is also different.

Now, let the coordinates of marks A' and B' on the screen be (xa, ya>, (xb, yb), respectively)
and the distortion correction coefficients Ex and εy of the optical system, the mark A at the camera coordinates (X', y'),
Coordinates of B (xa, ya), (xb, yb)
can be expressed as.

On the other hand, the coordinates of marks A and B in the factory coordinate system are ? -riA (Xa, Yb), ? -9B (Xb, Yb
) - (It is known in advance that it is IIE.

The factory coordinates of the origin P of the camera coordinates (x', yo) are p
(XO, yo), the relational expression for coordinate transformation is as follows, and by substituting the coordinate values of equations (13 and (II) above) into equation (III), 1, P (xo, yo) and θ are In addition, considering the distance (L) between the center of the camera and the center of the unmanned vehicle body, the position coordinates of the unmanned vehicle in the factory coordinate system and the direction from the center of the unmanned vehicle to the mounting of the camera are determined based on the direction of the center in the factory coordinate system. The angle of rotation to the left from the X axis is determined.

B) Concept of recognition of the current position and direction of movement of an unmanned vehicle using one mark recognition.

If one mark can be used to recognize the information about which mark it is and the information about the direction in which the mark is located (details will be described later), then 17
-The current location and direction of travel of the unmanned vehicle can be determined just by recognizing the location.

That is, although the above [■] equation is only one equation because there is one set of coordinate information, the value of θ is known from the above information on the direction, and the above P (xo, yo) can be found. The subsequent processing is the same as in case A) above.

C) Control structure of unmanned vehicle (see block diagram in Figure 4) Next, the control structure of the unmanned vehicle (1) according to the present application will be explained.

That is, this control structure is composed of the following blocks.

a) CCD camera <3) This camera (3) is oriented perpendicularly to the ceiling surface (4), and the image input through the lens is projected as image information on a screen composed of multiple pixels. Served.

b) Vision controller (10) Extracts only the target mark (5) from the image data of the ceiling surface captured by the above CCD camera (3), reads the center coordinates of the mark and the code of the mark on the CCD screen, and controls the main Periodically reports to the controller (11).

C) Main controller (11) A calculation unit (12), coordinate table (13), which will be explained below.
It is composed of a travel control section (14), map data (15), and a transfer control section (16).

d) Calculation 11B (12) From the local coordinate values for the mark (5) obtained from the vision controller (10), the mark code and the coordinate values of each mark in the factory coordinate system obtained from the mark coordinate table (13), calculate the above A). , B) to calculate the position coordinates and orientation of the unmanned vehicle (1) in the factory coordinate system, and periodically report these to the travel control unit (14).

e) Coordinate table (13) In the factory coordinate system for each mark (5) The coordinates are entered.

f) Travel control unit (14) Determines the travel route based on the destination and memory map data (15) instructed by the communication controller (17), and servo driver (18) which is the travel and steering drive system.
Instructs driving data.

Compare the position and direction data of the unmanned vehicle reported from the calculation unit (12) with the destination information from the map data (15), and if there is a discrepancy beyond tolerance, correct the driving data and drive. , servo driver (1) which is the steering drive system
8) Give instructions.

In addition to data on the position and direction of the unmanned vehicle (1) reported from the calculation unit (12), the encoder output (19) from the motor (M) of the driving/steering drive system is used to calculate its own position, Calculate the direction and interpolate the observation interval based on vision.

Data on the position, direction, and status of the unmanned vehicle (1) are periodically reported to the controller (17).

g) Communication controller (17) The communication unit receives destination, start, stop, transfer instructions, etc. from the ground controller (20) via the ground side communication controller (21) and communication unit (22).
3) and reports each time to the main controller (11).

Information such as the current point, next stop position, status code, etc. from the main controller (11) is sent to the ground controller (20).
Report in response to a call from.

h) Ground controller (20) Accepts transport requests from CPU (24), transport request setting board, etc., and assigns the transport request to the unmanned vehicle considered to be the most efficient.

Collision prevention is managed based on data such as the next stop reported by the unmanned vehicle (1), and instructions to stop and start are given to the unmanned vehicle (1) as necessary.

Various status data reported from the unmanned vehicle (1) is displayed in animation on the graphic display (25) to enable real-time monitoring of the system status.

In offline mode, use the CAD function to define maps and mark coordinates to create an unmanned vehicle (1
) has a data creation function to download to.

l) Transfer robot control unit (26) Although this embodiment shows a transfer port (27) mounted on an unmanned vehicle (1), the transfer robot (27) is A detailed explanation will be omitted since it deviates from the main purpose.

C) Marks Next, the marks will be explained in detail. Note that the marks described in detail below are related to the recognition of the current position and traveling direction of the unmanned vehicle by the two-mark recognition described in item A) above.

An example of mark (5) is shown in FIG. This mark (5) has a circular shape, and from its outer periphery, a first black sheet band (30) consisting of a black part all around the circumference, a first black sheet band (
a first white sheet band (31) consisting of the inner white part of 30);
The cord sheet band (3) inside the first white sheet band (31)
2), a second black sheet band (33) consisting of the black part inside the cord sheet band (32); a second white sheet band (34) consisting of the white part inside the second black sheet band (33); Center point seam h portion (
35). Each of the black sheet bands (30) (33) in the black area and the center point sheet part (35) are made of a sheet with a color that reflects less light, such as black, and the respective sheet bands (31) (34) are made of white, etc. Use a sheet that has a color that reflects a lot of light and is free from diffuse reflection, scratches, and stains.

Although the color of the ceiling surface may be either black or white, it is assumed that it is black for this embodiment. The code sheet band (32) is divided in the radial direction into n bits (7 bits in the figure, and an embodiment with 7 pits will be described below), and each bit is further divided into three equal parts. Hereinafter, the area divided into three equal parts will be referred to as a small area.

E) Operation of mark signal discrimination Next, the operation of distinguishing the mark (5) will be explained.

The image of the ceiling surface captured by the camera (3) is sent to the vision controller (10), and the controller (10)
Then, as shown in FIG. 12, the center position coordinates and code contents of the mark (5) are read out.

(First step) The surface information captured by the camera (3) is input into the controller (10) as a single signal by scanning in the Chinese direction. As an example, signals corresponding to the scanning lines (Scl to 5c5) in FIG. 5 are shown as signal lines (Sgl to 5g5) in FIG. 6, respectively.

(Second Step) Excessive brightness from a lighting fixture or the like is determined as a disturbance signal for the signal (details are omitted), and the signal is binarized by comparison with a set threshold value.

(Third Step) On the other hand, the horizontal tension (36) and vertical tension (37) as shown in FIG. 7.8 are previously input into the controller (10). The templates (36) and (37) have the same structure and correspond to the second white sheet portion (34) and the central black sheet portion (35) of the mark (5). That is, an arbitrary number of black detection pixel areas (hereinafter simply referred to as black pixel areas) on both sides (38) (39)
white pixel areas (40) each having the same width (Fa) as the width (Ha) of the second white sheet portion (31).
(41) is formed, and a black pixel area (42) having the same width (Fb) as the diameter of the central black sheet portion (35) is formed at the center inside thereof.

The above horizontal template (36) is compared with each image signal, and the horizontal coordinates of the matched locations are detected.

(Fourth step) At the same time as matching detection using the horizontal template (36), the vertical template (37)
Matching detection is performed using , and the vertical coordinates of the matched location are detected.

(Fifth step) By finding the intersection of the horizontal and vertical coordinates detected in the third and fourth steps, the center position coordinates (Sn) of the mark (5) within the camera (3) are detected.

(6th to 9th steps) As above, mark (5th to 9th steps)
) is detected, then the center position (Sn
Scanning is performed in the clockwise direction at a distance of radius (Ra) from >.

The scanning using the radius (Ra) is to scan exactly on the code sheet band (32).

The scanning start point can be anywhere, and the start bit (
St), and then reads the information on the code sheet band (32). Figure 9.10 shows the reading state.

In this example, the three subareas are black (b) and white (w).
- "0" for black (b), "0" for black (b) - white (W
) - White (W) corresponds to "1". Therefore, in the embodiment, the code "rololoo" is read, and the unmanned vehicle recognizes that the mark corresponds to the code. 2' for 6 bits
= 64 types, and in 10 bits, 2'' = 1024 types can be identified.

As described above, the mark (5) is circular and the code of the mark is also located on the circumference of the concentric circle. Therefore, regardless of the tilt of the unmanned vehicle, the center position of the mark and the code content can always be accurately read from any direction. If the last code among the above codes is "1" (there are five consecutive small areas with W>, there is a possibility that it may not be possible to recognize where the start bit is, so if three or more small areas are consecutive) , and then the point at which the black (b) small area appears is recognized as the start of the first bit.

Regarding the mark of the Flk embodiment, next, the mark related to the recognition of the current position and traveling direction of the unmanned vehicle by the one-mark recognition in item B) will be explained.

The mark is shown in FIG. In addition, the above D),
The same reference numerals are given to the same parts as in section E), and the explanation is omitted.

This mark (5) has a black II mark (43) attached to the first white sheet band (31) for direction recognition. Kurobe
43) is attached to all marks in the same direction and at the same position. For example, the black part (43) is attached due north from the mark center position (Sn).

The signal discrimination of the mark (5) is performed by the code sheet band (
32) Along with the above scanning, the first white sheet band (3
1) is also scanned to detect the position of the black part (43).

By detecting the direction of the black part (43) in the camera image, it is possible to relatively recognize the direction of the unmanned vehicle in the factory coordinate system. In this embodiment, the starting point of scanning is an arbitrary specific point, and the direction of the mark (5) is detected based on the distance from the specific point (starting point) to the black part (43). Note that it is also possible to use the start bit (St) as a substitute for the black part (43).

G) Regarding the guidance of unmanned vehicles, place the mark (5) described in item D) or item (YF) above on the camera (
3), the vision controller (10) reads the center position coordinates (Sn) and code contents of each mark (5), and based on the read information, the calculation unit (12) calculates the above item A) or B). Through this process, the current position and traveling direction of the unmanned vehicle (1) in the factory coordinate system are recognized. Based on the information recognized by the calculation unit (12), the travel control unit (14) instructs the travel data to the servo driver (18).
). The control unit (14) uses the map data (15) that stores the destination instructed by the communication controller (17) and the travel route within the factory, and the calculation unit (12).
) will guide the driving of the unmanned vehicle.

H) Layout changes If the travel routes stored in the above map data (15) are all routes within the factory, you can change the layout of unmanned vehicles by simply changing the destination instructions from the communication controller (I7). The run will be changed. In addition, if you change the layout in the factory and make a new driving route from a part that was not a driving route, or vice versa, please use the map data (15).
All you need to do is change some of the memory contents. Of course, the mark (5) on the ceiling and the camera (3) on the unmanned vehicle are the squares.

In other words, since the unmanned vehicle can recognize its current location as the factory coordinates at any location within the factory, when changing the driving route, at least the coordinate data given to the unmanned vehicle only needs to be changed. , no other hard changes are required.

〔Effect of the invention〕

As is clear from the above description, according to the present invention, an unmanned vehicle is guided without the need for complex image processing, and the unmanned vehicle is constantly aware of its current position while driving. Therefore, it has the flexibility to respond immediately when changing the layout, making it more intelligent.

[Brief explanation of the drawing]

Figure 1 is a schematic diagram that schematically shows the operation status of an unmanned vehicle inside a building, Figure 2 is a schematic diagram to explain the concept for position recognition of an unmanned vehicle, and Figure 3 is a side view as well. , FIG. 4 is a block diagram showing the control structure of the unmanned vehicle, FIG. 5 is a front view showing the mark, FIG. 6 is a schematic diagram for explaining the scanning state of the mark, and FIG. 7 is a horizontal template. Schematic diagram: Fig. 8 is a schematic diagram showing a vertical template; Fig. 9 is a schematic diagram illustrating scanning for reading mark code information; Fig. 10 is a schematic diagram illustrating analysis of code information. 11 is a front view showing another embodiment of the mark, and FIG. 12 is a flow chart for mark signal discrimination. (1) Unmanned vehicle (2) Building (3) Camera (imaging means) (4) Ceiling surface (5) Mark (10) Vision controller (analysis means) (12
) Calculation unit (calculation means) (13) Coordinate table (storage means) (14) Travel control unit (15) Manob data (32) Code sheet band (circular code) Fig. 6

Claims (1)

[Claims] Distinguishable marks are arranged at predetermined positions on the ceiling of a building into which unmanned vehicles are installed, and the unmanned vehicles are equipped with means for photographing the ceiling and for analyzing information on the photographed images. means for storing the intra-building coordinates of each mark; and means for calculating the current position and orientation of the unmanned vehicle by inputting the analyzed image information and the intra-building coordinates of the marks in the storage means;
A guidance device for an unmanned vehicle, comprising: means for controlling the traveling of the unmanned vehicle based on the output of the calculation means.