JP2010102585A - Mobile object position detection system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To detect the position of a mobile object using a pattern drawn on a plane. <P>SOLUTION: The mobile object position detection system includes: a plane configured by arranging a plurality of panels adjacent to each other with the same dot pattern arranged thereon; a reference feature amount storage means for calculating the reference feature amount of each dot configuring the dot pattern of the plurality of panels and storing the reference feature amount in a database in association with a coordinate value on the plane of each dot; an imaging means for imaging an area on the plane where the mobile object is positioned at predetermined time intervals; a detection feature amount calculation means for selecting a predetermined one of the plurality of dots in the image acquired by the imaging means and calculating the detection feature amount of the selected dot; and a feature amount collation means for selecting the reference feature amount of the highest matching degree by collating the detection feature amount with the reference feature amounts in the collation range of a predetermined size in the database, and detecting the coordinate value of the selected reference feature amount as a present position of the mobile object on the plane. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a system for detecting the position of an object moving on a plane.

  Conventionally, methods such as a ceiling camera method, a ceiling ID installation method, an ultrasonic wave utilization method, and an RFID utilization method are known as methods for detecting the position of a moving body such as a transport machine, a robot, and a human in a building.

  The ceiling camera system is a method of determining a position by imaging a mark installed on an object using a camera installed on a ceiling or a wall and detecting the mark. The ceiling ID installation method detects the position and direction of the camera by installing an ID marker that reflects infrared rays on the ceiling, irradiating infrared rays from below and capturing the reflection with the camera. However, these methods cannot be established unless a structure such as a ceiling or a wall is located above. Further, when there is an object that blocks between the camera and the mark, the detection becomes impossible.

  In the ultrasonic usage method, multiple ultrasonic microphones are placed on the ceiling or wall, an ultrasonic speaker is installed on the object, and the time from when the sound generation command is sent to the ultrasonic speaker until the sound is heard by the ultrasonic microphone. Is measured, the distance between the ultrasonic speaker and the plurality of ultrasonic microphones is determined from this time, and the three-dimensional position of the speaker is calculated by triangulation. However, in this method, if an object enters between the speaker and the microphone, the sound is cut off and measurement is impossible. In addition, a large measurement error occurs due to reflection of ultrasonic waves by walls and columns.

  The RFID system is a method in which a large number of RFID tags are arranged on a floor, these RFID tags are received by an RFID receiver, and a detection position is specified by the ID of the received RFID tag. However, generally, since RFID tags are detected discontinuously, the position cannot be detected at a position where the RFID cannot be detected. Although it is possible to detect near-continuous tags by spreading RFID tags at a high density, the cost increases and the detection values are discrete in units of several centimeters, and the position can be located only with a resolution of several centimeters at the minimum. It cannot be specified.

  Furthermore, the method of detecting the position of the moving body moving on the plane using a pattern (pattern) arranged on the plane is not limited to the position detection of the moving body in the building as in the above various methods. It is known (see, for example, Patent Document 1).

  In the position detection system of Patent Document 1, a coding pattern obtained by coding a plane XY coordinate value is arranged on a plane. This coding pattern includes a plurality of grid-like reference lines, and a plurality of dots are arranged in association with the positions of the respective intersections of the reference lines. The position detection device provided in the moving body captures an image of an area on a plane where the moving body is located, processes the acquired image to identify the coding pattern, and restores the coding pattern to restore the moving body. An XY coordinate value on the plane is obtained.

JP 2006-141061 A

  As shown in Patent Document 1, in a position detection system that detects a position of a moving body that moves on a plane using a pattern arranged on the plane, one encoding pattern is assigned to each coordinate value on the plane. Since it corresponds, there exists a problem that the freedom degree of design of a pattern is low. In particular, in the position detection system of Patent Document 1, since dots are arranged around the intersection of the grid-like reference line, only a pattern close to the grid pattern can be applied.

  In view of the above points, the present invention detects the position of a moving body using a pattern drawn on a plane without using a pattern encoded in correspondence with the coordinate value of the plane as in the prior art. It is an object to provide a movable body position detection system capable of

  The moving body position detection system according to the present invention includes a plane configured by arranging a plurality of panels having the same dot pattern arranged on the surface thereof, individual dots constituting the dot pattern, and individual dots. A positional relationship with a plurality of dots located in the vicinity is calculated as a reference feature amount of each individual dot, and the reference feature amount is associated with a coordinate value on the plane of each individual dot in a reference feature amount database. Reference feature quantity storage means for storing, an imaging means provided in a moving body that moves on the plane, and that captures an area on the plane on which the moving body is located at a predetermined time interval; and A predetermined dot is selected from a plurality of dots in the acquired image, and the positional relationship between the selected dot and a plurality of dots positioned around this dot is determined based on the selected dot. A detected feature amount calculating means for calculating as an output feature amount, and the detected feature amount most closely matches the detected feature amount by comparing the detected feature amount with a reference feature amount of a matching range of a predetermined size in the reference feature amount database. And a feature amount matching unit that selects a reference feature amount having a high degree and detects a coordinate value of the selected reference feature amount on the plane as a current position of the moving object on the plane.

  A moving body position detection system according to claim 2 of the present invention is the moving object position detection system according to claim 1, wherein the dot pattern is obtained by dividing the panel into a plurality of blocks and arranging dots at random positions in each block. It has been created.

  The moving body position detection system according to claim 3 of the present invention is the moving body position detection system according to claim 1 or 2, wherein the reference feature amount and the detected feature amount are within a region of a dot and a circle having a predetermined radius centered on the dot. It is characterized by including an array composed of the direction of each line segment connecting a plurality of positioned peripheral dots and an array composed of the length of each line segment.

  According to a fourth aspect of the present invention, in the mobile body position detection system according to any one of the first to third aspects, the collation range is set to a range that does not include dots having the same reference feature amount. It is characterized by that.

  According to a fifth aspect of the present invention, in the moving body position detection system according to any one of the first to third aspects, the collation range includes a maximum moving speed of the moving body and an image acquisition time interval of the imaging means. It is set to the peripheral area | region which includes the movable range of the said mobile body determined from (1).

  According to the movable body position detection system of the present invention, the design of the dot pattern arranged on the plane can be freely compared with the conventional case where detection is performed using the dot pattern encoded corresponding to the plane coordinate value. The degree can be improved.

  In addition, since the conventional position detection system has a different dot pattern on the entire floor surface, when constructing the floor surface with materials such as multiple panels and tiles, lay out the panels in the correct order and orientation. However, according to the moving body position detection system of the present invention, the floor surface is configured by arranging a plurality of panels having the same dot pattern in parallel. By this, the workability at the time of installing a panel can be improved.

  Exemplary embodiments of a mobile body position detection system according to the present invention will be described below in detail with reference to the accompanying drawings. Below, the example which the mobile body which moves on the floor surface in a building detects an own position is demonstrated.

(Embodiment 1)
FIG. 1 is a conceptual diagram of a moving body position detection system 1 according to the present embodiment, and the left figure of FIG. 2 shows a dot pattern arranged on a floor F (floor panel f) that is a moving area of the moving body. 3 is a diagram illustrating an example of an image obtained by capturing a part of the dot pattern 3. FIG. FIG. 3 is a plan view of the floor F, and FIG. 4 is a conceptual diagram of a collation process in the moving body position detection system 1.

  A moving body position detection system 1 illustrated in FIG. 1 is provided on a floor F configured by arranging a plurality of panels in parallel and a moving body (not shown), and moves on the floor F together with the moving body. It is comprised from the moving body position detection apparatus 10. FIG. Here, the moving body is, for example, a transport carriage, a robot, or a human.

As shown in FIG. 3, the floor F is composed of a plurality of floor panels f _{11 to} f ₃₃ formed in a square shape having the same size. The plurality of floor panels f _{11 to} f ₃₃ are juxtaposed in the X direction and the Y direction in a state where adjacent sides are in contact with each other. Here, the two-digit subscripts of the floor panels f _{11 to} f ₃₃ indicate the alignment numbers of the floor panel f in the X direction and the Y direction.

As shown in FIG. 3, the same dot pattern 3 is arranged on the surface of each floor panel f _{11 to} f ₃₃ . That is, the entire floor F is a repetition of the same dot pattern 3.

  The moving body position detection apparatus 10 captures an area on the floor F where the moving body is located at predetermined time intervals, and processes the image captured by the imaging section 11 to process the position of the moving body. And a control unit 15 that performs detection.

As shown in FIG. 4, the moving position detecting device 10, before performing the position detection of the moving object, previously calculated feature amounts of all the dots 5 constituting the dot pattern 3 of the floor panel f ₁₁ as a reference This is stored in the reference feature database 16. As will be described later, this feature amount is obtained by calculating the positional relationship between each dot 5 in the dot pattern 3 and a plurality of dots 5 positioned around each dot 5. Hereinafter, this is referred to as “reference feature amount”. The calculated reference feature value of each dot 5 is associated with the absolute coordinate value on the floor F of each dot 5 and stored in the reference feature value database 16.

  Further, when the moving body position detecting device 10 moves together with the moving body, the moving body position detecting device 10 images a region on the floor F where the moving body position detection apparatus 10 is located, and a predetermined dot 5A (see FIG. 4) from a plurality of dots 5 in the acquired image. ) And the feature amount of the dot 5A is calculated. As will be described later, this feature amount is obtained by calculating the positional relationship between a dot 5A in the image and a plurality of dots 5B located around the dot 5A. Hereinafter, this is referred to as “detected feature amount”.

  Then, the moving body detection device 10 collates the detected feature amount of the dot 5A in the acquired image with the reference feature amount of the collation range of a predetermined size in the reference feature amount database 16, thereby obtaining the highest degree of coincidence. A reference feature value is selected, and the coordinate value of the selected reference feature value is detected as the current position of the moving object on the floor F.

[Create dot pattern]
FIGS. 5A and 5B are diagrams for explaining an example of a method of creating the dot pattern 3 on the surface of the floor panels f _{11 to} f ₃₃ . Figure 5-1 shows an enlarged part of the floor panel f ₁₁ as a reference. As described above, since the floor panels f _{11 to} f ₃₃ constituting the floor F have the same dot pattern 3, the calculation described below is performed only on one floor panel f _{11 serving as} a reference. Just do it.

  The plurality of dots 5 constituting the dot pattern 3 are in principle randomly printed. However, if the dots 5 are completely randomly shot, the minimum number of dots 5 necessary for the imaging region of the imaging unit 11 is required. It can happen that there is no. In such a case, since it becomes impossible to detect the position of the moving body, the floor surface is such that the minimum required number of dots 5 are present in the image no matter which region on the floor surface 2 is imaged. It is necessary to hit the dots 5 uniformly on 2.

Considering the above points, in the present embodiment, as shown in FIG. 5A, the surface of the floor panel f ₁₁ is divided into a plurality of square blocks (regions) 4, and random in each block 4. A dot pattern 3 is created by hitting one dot 5 at a position. Various methods are conceivable for randomly placing dots 5 in each block 4, but in this embodiment, each block 4 is subdivided into 4 N square small blocks, and each block 4 A method is adopted in which a numerical value is randomly generated corresponding to the above, one small block is selected based on this numerical value, and the dot 5 is assigned to the center of the selected small block.

For example, as the simplest example, as shown in FIG. 5B, the block 4 is divided into four (four to the power of 4) square small blocks 4a to 4d, and the randomly generated numerical value λ _ij is used. Then, the dot 5 is hit on any one of the small blocks 4a to 4d. Here, i and j in λ _ij are numbers indicating the position of the block 4, i is a number in the block vertical direction, and j is a number in the block horizontal direction. As shown in FIG. 5B, if the range is 0 <λ _ij ≦ 0.25, the dot 5 is hit on the small block 4a. If the range is 0.25 <λ _ij ≦ 0.50, the dot 5 is hit on the small block 4b. If the range is 0.50 <λ _ij ≦ 0.75, the dot 5 is hit on the small block 4c. If the range is 0.75 <λ _ij ≦ 1.0, the dot 5 is hit on the small block 4d. For example, as shown in FIG. 5A, since λ ₁₁ is 0.78, dot 5 is hit on the small block 4d, and since λ ₁₂ is 0.32, dot 5 is hit on the small block 4a. . This By performing all blocks 4 on the floor panel f _11, it is possible to create a dot pattern 3 on the floor panel f _11. The above-described method for creating the dot pattern 3 is an example, and various other creation methods are conceivable.

[Create reference feature database]
Next, a method for calculating the reference feature amount of each dot 5 in the dot pattern 3 will be described. As described above, the floor panels f _{11 to} f ₃₃ have the same dot pattern 3. Therefore, the calculation of the reference feature value described below is performed only for one of the floor panel f ₁₁ as a reference.

As shown in FIG. 6, a circle having a predetermined radius centered on the dot 5a to be calculated is assumed. Hereinafter, the dot 5a is referred to as “center dot 5a”. The radius of this circle is not particularly limited, but in this embodiment, the radius of the circle is 1.25 times the length of one side of the block 4. In the example shown in FIG. 6, there are five peripheral dots 5 b _{1 to 5} b 5 in a circle area centered on the center dot 5 a.

Next, draw a five arms connecting the center dot 5a and the peripheral dots _5b 1 _{~5b 5} (segment) to select based on the one of the arms of the arms 6a to 6e, the arm and the other 4 The direction of each arm 6a-6e is calculated | required from the angle which the arm makes. Further, the lengths of the five arms 6a to 6e are obtained.

Positional relationship between the center dots 5a and the peripheral dots _5b 1 _{~5b 5} is composed of a number of arms 6a to 6e, the arrangement direction of each arm 6a to 6e, the sequence length of each arm 6a to 6e . Taking the central dot 5a shown in FIG. 6 as an example, the number of arms is 5, and the arrangement in the direction of each arm 6a to 6e is (0, 90, 150, 240, 300) with respect to the arm 6a, and each arm 6a. The array having a length of ˜6e is represented as (1, 0.5, 1.2, 0.5, 1.2), where the length of one side of the block 4 is 1. Hereinafter, the arrangement in the direction of the arm is omitted and referred to as “arm direction arrangement”, and the arrangement of the length of the arm is omitted and referred to as “arm length arrangement”.

The number of arms, the arm direction arrangement, and the arm length arrangement described above are calculated for all dots 5 of the dot pattern 3. This is the reference feature amount of each dot 5 of the dot pattern 3. The reference feature quantity of each dot 5 of the floor panel f ₁₁ is stored in the reference feature quantity database 16 in association with the absolute coordinate value of the floor F of each dot 5.

For the other floor panels f _{12 to} f ₃₃ , relative coordinates with respect to the reference floor panel f ₁₁ (where the floor panels f _{12 to} f ₃₃ are arranged relative to the reference floor panel f ₁₁ are shown. Are stored in the reference feature database 16. As will be described later, when the position of the moving body is detected, if the relative positions of the other floor panels f _{12 to} f ₃₃ with respect to the reference floor panel f ₁₁ are known, the other floor panel f ₁₂ is detected. You can obtain the absolute coordinates of each dot in ~f _33.

[Moving object position detection processing]
Next, the process in which the moving body position detection apparatus 10 detects the position of the moving body on the floor 2 will be described. FIG. 7 is a block diagram of the moving body position detection apparatus 10. As shown in FIG. 7, the moving body position detection apparatus 10 includes an imaging unit 11, a dot pattern creation unit 12, a reference feature amount calculation unit 13, a reference feature amount storage unit 14, a control unit 15, and a storage unit 19. ing.

  The imaging unit 11 is an image acquisition device such as a CCD camera. The imaging unit 11 is fixed to a moving body, and while moving with the moving body, captures an area on the floor surface 2 where the moving body is located at a predetermined time interval, and acquires an image as shown in FIG. . In the present embodiment, images are acquired at a time interval of about 10 ms, for example, using a high-speed camera. Moreover, as shown in FIG. 1, the distance between the imaging part 11 and the floor surface 2 is several centimeters, and the image with little influence of disturbance light is acquirable by using the illumination 11a.

  The dot pattern creation unit 12 calculates the position of each dot 5 arranged on the floor panel f, and creates the dot pattern 3 on the floor panel f. The procedure for creating the dot pattern 3 is as described above.

  The reference feature value calculation unit 13 calculates the reference feature values of all the dots 5 in the dot pattern 3 of the floor panel f created by the dot pattern creation unit 12. The calculation procedure of the reference feature amount is as described above.

Reference characteristic amount storage unit 14, a reference feature amount of individual dots 5 of the floor panel f ₁₁ as a reference calculated by the reference feature amount calculating section 13 stores in association with the absolute coordinate values of the floor F, the reference The relative coordinate values of the other floor panels f _{12 to} f ₃₃ with respect to the floor panel f ₁₁ are stored, and has the above-described reference feature quantity database 16.

  The control unit 15 detects the position of the moving body on the floor 2 by processing the image picked up by the image pickup unit 11, and includes a detection feature amount calculation unit 17 and a feature amount comparison unit 18. Yes.

  As shown in FIG. 4, the detection feature amount calculation unit 17 uses a plurality of dots 5 in the image acquired by the imaging unit 11 to form a dot 5A (hereinafter referred to as a “center dot 5A”) located near the center of the image. And the positional relationship between the selected dot 5A and a plurality of dots 5B located around the dot 5A is calculated as a detected feature amount of the dot 5A.

The detected feature amount is calculated in the same manner as the reference feature amount calculation method described above. In the acquired image shown in FIG. 4, a circle having a predetermined radius with the center dot 5A as the center is assumed. The radius of this circle is the same as the radius of the circle when the reference feature value is calculated. In the example shown in FIG. 4, there are five peripheral dots 5 </ b> B _{1 to 5} </ b> B 5 in a circular area centered on the central dot 5 </ b> A. Draw _five arms 6A to 6E that connect the center dot 5A and the peripheral dots 5B _{1 to} 5B ₅ and select any one of the five arms 6A to 6E as a reference. The direction of each arm is determined from the angles θ _{1 to} θ ₅ (θ ₁ is 0 °) formed by the four arms. Further, the lengths L _{1 to} L ₅ of the _five arms 6A to 6E are obtained. That is, the detected feature amount of the center dot 5A includes the number of arms 6A to 6E, the arm direction array (θ ₁ , θ ₂ , θ ₃ , θ ₄ , θ ₅ ), and the arm length array (L ₁ , L ₂ , L _3, represented de _L 4, _{L 5)} and.

  The feature amount matching unit 18 matches the detected feature amount of the center dot 5A in the acquired image calculated by the detected feature amount calculating unit 17 with the reference feature amount in the matching range of a predetermined size in the reference feature amount database 16. Thus, the reference feature amount having the highest degree of coincidence with the detected feature amount of the center dot 5A is selected. Then, the absolute coordinate value of the selected dot is detected as the current position of the moving object on the floor F. In the following, the reference feature quantity in the reference feature quantity database 16 within the collation range of a predetermined size is referred to as “reference feature quantity data list 20”.

  As described above, the entire floor F is a repetition of the same dot pattern 3. Therefore, when the detected feature quantity of the center dot 5A in the image is collated with the reference feature quantity data list 20, if the size (collation range) of the reference feature quantity data list 20 is excessively large, the collation range is included. A situation occurs in which two or more dots 5 having the same reference feature amount appear. That is, dots 5 at the same position on different floor panels are included in the collation range. In such a case, the position of the moving body cannot be detected. The standard of the size of the collation range in which dots having the same reference feature amount are included in the collation range is about the size of one floor panel. For this reason, the above collation range must be made smaller. The size of the collation range can be determined as follows.

FIG. 8 conceptually shows the trajectory of the moving body moving on the floor F. Points P _{1 to} P _n in FIG. 8 respectively represent the position of the moving body at the time of image acquisition by the imaging unit 11, that is, the detection position of the moving body. The imaging unit 11 acquires images at regular time intervals. Point P ₁ is the initial position of the moving body (movement start position). An in-circle region C ₁ having a predetermined radius with the initial position P ₁ as the center is the above-described collation range. The collation range C _1, specifically, means a range for searching the position P ₂ of the mobile during the next image acquisition.

As shown in FIG. 8, collation range C ₁ is a peripheral region containing the movable range M ₁ from the initial position P ₁ of the moving body. Here, the movable range M ₁ of the moving body is a moving range of the moving body determined from the maximum moving speed of the moving body and the image acquisition time interval of the imaging unit 11. To explain with a specific example, if the image acquisition interval of the imaging unit 11 is 10 ms (1/100 second) and the maximum moving speed of the moving body is 1.0 m / s, the image acquisition interval of 10 ms for one time In between, the moving body advances 10 mm. In this case, the movable range M ₁ of the moving body is within a circular area having a radius of 10 mm with the moving body as the center. In other words, the movable range is a range in which the moving body can move between the time when the imaging unit 11 acquires an image and the time when the next image is acquired. Therefore, the collation range C ₁ needs to be set larger than the movable range M ₁ , but as shown in FIG. 8, if it is set to a peripheral region that is slightly larger than the movable range M ₁ , the position P ₂ of the mobile body at the time of image acquisition can be reliably detected.

However, as described above, collation range C ₁ must be within a range that does not include dots 5 have the same reference feature amount. Therefore, when the movable range M1 shown in FIG. 8 exceeds the width of _one floor panel, the position cannot be detected. In such a case, it sets a short image acquisition time interval of the imaging unit 11, should be smaller than the movable range M ₁ of one floor panel size.

  The storage unit 19 is a part that temporarily stores the reference feature value data list 20 extracted from the reference feature value database 16.

  FIGS. 9-12 is a flowchart which shows the flow of the process which the control part 15 of the moving body position detection apparatus 10 comprised as mentioned above performs. FIG. 9 is a main flowchart, and FIGS. 10 and 11 are sub-flowcharts of FIG. FIG. 12 is a sub-flowchart of FIGS. 10 and 11. Hereinafter, a processing procedure executed by the control unit 15 will be described with reference to FIGS. 8 and 9 to 12.

Before performing the detection process, the array number of the floor panel from which the moving body starts moving is input in advance. In the example shown in FIG. 8 example, since the initial position P ₁ of the moving body is on the floor panel f _13, enter "f _13". Control unit 50 of the mobile position detection device 10, when the movement start panel f ₁₃ is inputted (step S1), and detection processing of the initial position P ₁ of the moving body. Procedure for detecting the initial position _{P 1} of the movable body, the procedure SUB1 shown in FIG. 10, there are two steps SUB2 shown in FIG. 11. First, the procedure shown in FIG. 10 will be described.

As shown in FIG. 10, the control unit 15 acquires the image data of the initial position _{P 1} of the moving object from the imaging unit 11 (SUB1: step S11), and the one of the plurality of dots of the image dot 5A (center dot 5A) is selected, and the detected feature value of the center dot 5A is calculated through the detected feature value calculation unit 17 (SUB1: step S12). Here, as shown in FIG. 4, the detected feature amount is the number of arms connecting the central dot 5A and the plurality of peripheral dots 5B, the arm direction arrangement, and the arm length arrangement. Control unit 15, when the arms the number of center dot 5A is equal to or greater than the threshold (SUB1: Step S13: Yes), then extracted reference feature amount data list 20 from the reference feature quantity database 16 corresponds to the movement start panel _{f 13} This is stored in the storage unit 19. And the control part 50 collates the detection feature-value of center dot 5A with the reference | standard feature-value data list 20 of the memory | storage part 19 through the feature-value collation part 18 (SUB3). On the other hand, when the number of arms in the center dot 5A is less than the threshold (SUB1: Step S13: No), another dot in the image is newly selected as the center dot 5A without collating with the reference feature amount data list 20. (SUB1: Step S14), the process returns to Step S12 to calculate the detected feature amount of the newly selected center dot 5A. Here, the threshold value of the number of arms in step 13 is appropriately set according to the density of the dots 5 of the dot pattern 3 or the like. In this embodiment, the number of arms is four.

  The reason why the threshold value is provided for the number of arms of the detected feature amount in step S13 is that, when the number of arms is small, the rate at which the collation result is determined as one decreases, and the purpose is to speed up the calculation. However, step S13 is not necessarily provided, and step S13 may be omitted.

  The process (SUB3) for collating the detected feature quantity of the center dot 5A in the image with the reference feature quantity data list 20 is performed according to the procedure shown in FIG. First, a dot that matches the arrangement in the arm direction of the center dot 5A in the image is selected from the reference feature value data list 20 (SUB3: Step S41).

  As described above, in the reference feature value data list 20 (reference feature value database 16), each dot is arranged in one direction in the direction of the arm and the angle between a specific reference arm and another arm. Stored. Therefore, there is a possibility that the reference arm in the arm direction arrangement of the central dots 5A in the image and the reference arm in the dot arm arrangement of the reference feature quantity data list 20 do not necessarily match. Therefore, an array using each arm as a reference arm is created as an array in the arm direction of the center dot 5A in the image. That is, when the number of arms is five like the central dot 5A in the image shown in FIG. 4, five types of arm direction arrangements are made with the five arms as reference arms, and the five types of arm directions are formed. The arrangement is checked against the arrangement in the arm direction of the reference feature value data list 20.

When the arm direction arrangements match, the degree of coincidence between the matched arm length arrangement of the reference feature quantity data list 20 and the center dot arrangement of the center dot 5A in the image is obtained (SUB3: step S42). As described above, the arm length array of each dot in the reference feature value data list 20 stores one kind of array in which the arm lengths are arranged in order from a specific reference arm. For this reason, when obtaining the degree of coincidence of the arm length arrays, an array in which the order of the arm lengths of the center dots 5A in the image is rearranged is also created. That is, when the number of arms is five like the central dot 5A in the image shown in FIG. 4, (L ₁ , L ₂ , L ₃ , L ₄ , L ₅ ), (L ₂ , L ₃ , L ₄ , L ₅ , L ₁ ), (L ₃ , L ₄ , L ₅ , L ₁ , L ₂ ), (L ₄ , L ₅ , L ₁ , L ₂ , L ₃ ), (L ₅ , L ₁ , L ₂ , L ₃ , L ₄ ), and the degree of coincidence between the five types of arm length arrays and the arm length array of the reference feature quantity data list 20 is obtained. Then, the one with the highest degree of matching is selected from these (SUB3: Step S43).

  As an example of a method for obtaining the degree of coincidence of arm length arrays, there is a method described below. The above five types of arm length arrays are divided by the size of each array to form a unit vector. Also, the arm length array of the reference feature amount is divided by the array size into a unit vector. Then, the inner product of each of the five types of arm length arrays (unit vectors) and the reference feature amount arm length array (unit vector) is calculated, and the value of the inner products is used as the degree of coincidence. That is, the one with the inner product value closest to 1 is assumed to have the highest degree of coincidence. Note that the use of the inner product value as the matching degree of the arm length array is merely an example, and other methods can be used.

  Note that the collation of the arm direction arrangement and the reference for the degree of coincidence of the arm length arrangements have an allowable range for errors caused by image resolution and the like.

  After the reference feature database matching process SUB3 is completed, as shown in FIG. 10, the control unit 15 determines whether there is one that matches the center dot 5A in the image (SUB1: step). S15). If there is only one that matches (SUB1: Step S15: Yes), the absolute coordinate value on the floor F of the matching dot in the reference feature data list 20 is detected as the current position.

The absolute coordinate value of the matched dot 5 can be obtained as follows. As described above, the reference feature amount data list 20 stores the relative coordinates of the other floor panels f _{12 to} f ₃₃ with respect to the reference floor panel f ₁₁ . In panel f ₁₁ as a reference, a dot having the same feature amount as matched dots of the reference feature amount data list 20 as P _{n (not} shown), the absolute coordinate values of the P _n, relative to the floor panel f ₁₃ If the coordinates are added, the absolute coordinate value of the matched dot is obtained. This absolute coordinate value is set as the current position of the initial position P1 (SUB1: Step S16).

  On the other hand, if there are multiple matches (SUB1: Step S15: No), the matching result is not adopted, and another dot 5 in the image is newly selected as the center dot 5A (SUB1: Step S14). Then, the detected feature amount of the newly selected center dot 5A is calculated (SUB1: step S12). Thereafter, the processes from step S12 to step S15 are repeated until the collation result is finally determined as one.

When detecting the initial position P ₁ of the moving body, the control unit 15 extracts the reference feature quantity data list 20 of the dots in the collation range C ₁ centered on the initial position P ₁ from the reference feature quantity database 16, and this is extracted. Store in the storage unit 19 (step S2). And the detection process of the present position of a moving body is started (SUB1).

  When acquiring image data from the imaging unit 11 (SUB1: step S11), the control unit 15 selects one dot (center dot 5A) from a plurality of dots in the image, and through the detected feature amount calculation unit 17, the center dot is selected. The detection feature amount of 5A is calculated (SUB1: Step S12). When the number of arms in the center dot 5A is equal to or greater than the threshold (SUB1: Step S13: Yes), the control unit 15 sends the detected feature amount of the center dot 5A to the reference feature amount data in the storage unit 19 through the feature amount matching unit 18. The list 20 is checked (SUB3). On the other hand, when the number of arms in the center dot 5A is less than the threshold (SUB1: Step S13: No), another dot 5 in the image is newly selected as the center dot 5A without collating with the reference feature amount data list 20. (SUB1: Step S14), the process returns to Step S12 to calculate the detected feature amount of the newly selected center dot 5A.

Control unit 15, in step S41~ step S43 in SUB3 collates the reference feature amount data list 20 in the comparison range C _1, that coincides with the center dot 5A in the image is equal to or one (SUB1: Step S15). If a match was was one (SUB1: Step S15: Yes), the absolute coordinate value on the floor F of the matched dots of the reference feature amount data list 20, as the current position P ₂ of the mobile Detect (SUB1: Step S16). Method of determining the absolute coordinates of the matched dots 5 are the same as for the initial position P ₁ above.

Control unit 15, after detecting the current position P ₂ of the mobile returns to step S2, the reference feature quantity database 16, extracts the reference feature amount data list 20 for the next comparison range C _2, which storage unit 19 is stored. As described above, the control unit 15 thereafter repeats the detection process of the current position (P ₃ , P ₄ ,..., P _n ,...) Of the moving body.

As in the collation range _{C 3} shown in FIG. 8, in a case where the collation range floor panel _{f 13,} the floor panel _{f 12,} the floor panel _{f 22,} spans a plurality floor panels of the floor panels _{f 23} Also, the reference feature quantity of dot 5 in each verification range of the floor panel f ₁₃ , the floor panel f ₁₂ , the floor panel f ₂₂ , and the floor panel f ₂₃ is extracted from the relative coordinates of each floor panel with respect to the reference floor panel f ₁₁ . can do. The control unit 15 collates with the reference feature quantity of the dot 5 in each area of the floor panel f ₁₃ , floor panel f ₁₂ , floor panel f ₂₂ , floor panel f _{23 in} the collation range C ₃ , and determines the position of the current position P ₄ . To detect.

Next, the case where the procedure (SUB2) shown in FIG. 11 is adopted as a procedure for detecting the current position of the moving body will be described. In the following described procedure for detecting the initial position P ₁ of the moving body.

As shown in FIG. 11, the control unit 15 acquires the image data of the initial position P ₁ of the moving object from the imaging unit 11 (SUB2: step S21), and the dots 5 first one of a plurality of dots in an image ( The first center dot 5A ₁ ) is selected, and the detected feature amount of the first center dot 5A ₁ is calculated through the detected feature amount calculation unit 17 (SUB2: step S22). Control unit 15, when the arm number in the first center dot 5A ₁ is equal to or greater than the threshold (SUB2: Step S23: Yes), the reference feature amount data list 20 corresponding to the movement start panel _{f 13} from the reference feature quantity database 16 Is extracted and stored in the storage unit 19. Then, through the feature checker 18 collates the detection characteristic of the first center dot 5A ₁ and the reference feature amount data list 20 in the storage unit 19 (SUB3). On the other hand, when the arm number in the first center dot 5A ₁ is less than the threshold (SUB2: Step S23: No), without checking with the reference feature amount data list 20, a new first central another dot in the image select a dot 5A ₁ (SUB2: step S24), and newly calculates the detected feature amount of the first central dot 5A ₁ selected returns to step S22.

Performs verification processing of steps S41~S43 of SUB3 shown in FIG. 12, the reference feature amount data list 20, it selects the first center dot 5A ₁ the highest degree of coincidence dots in the image. When a plurality of dots are selected, these are set as a “matching candidate group” (SUB2: step S25).

Next, the control unit 15 selects a _second dot (second center dot 5A ₂ ) from the plurality of dots in the image, and detects the detected feature amount of the second center dot 5A ₂ through the detection feature amount calculation unit 17. Calculate (SUB2: Step S26). Control unit 15, when the arm number of second central dot 5A ₂ is equal to or greater than the threshold (SUB2: Step S27: Yes), through the feature checker 18, a storage unit to detect feature amount of the second central dot 5A ₂ The 19 reference feature data list 20 is collated (SUB3). On the other hand, when the arm number of second central dot 5A ₂ is less than the threshold (SUB2: Step S27: No), without checking with the reference feature amount data list 20, the second another dot 5 in the image newly select the center dot 5A ₂ (step S28), newly calculates the selected detection feature quantity of the second central dot 5A ₂ returns to step S26.

Performs verification processing of steps S41~S43 of SUB3 shown in FIG. 12, the reference feature amount data list 20, and selects the second center dot 5A ₂ the highest degree of coincidence dots in the image. When a plurality of dots are selected, these are set as a “matching candidate group” (SUB2: step S29).

Then, the control unit 15, a first matching candidate group of the center dot 5A ₁ in step S25, the combination of the second matching candidate group of the center dot 5A ₂ in step S29, selects the appropriate combination (SUB2: Step S30). Specifically, the distance between the two dots, the equal to the first center dot 5A ₁ in the acquired image and the distance between the second center dot 5A _2, is selected as the optimum combination.

Then, the control unit 15, the absolute coordinate value on the floor F of the midpoint of the two dots selected from each match candidate set as the optimal combination is detected as the initial position P ₁ of the moving body (SUB2: step S31) .

In the processing of SUB1 described above, until the matching result of the reference feature value data list 20 is finally determined to be one, the calculation of the detected feature value and the matching process are repeated by changing the center dot 5A in the image. In the SUB2 process described later, it is not necessary to perform these processes many times, and the current position can be specified by a single process. Therefore, the SUB2 process has an advantage that the detection time is faster than the SUB1 process. Incidentally, for the next detection position P ₂ and subsequent location processing, similarly to the to perform processing of SUB2.

  As described above, the moving body position detection system 1 according to the present embodiment has dots with the floor F configured by arranging a plurality of floor panels f having the same dot pattern 3 arranged on the surface. The positional relationship between the individual dots 5 constituting the pattern 3 and the plurality of dots 5 positioned around the individual dots 5 is calculated as a reference feature amount of each dot 5, and the reference feature amount is calculated for each individual dot 5. A reference feature quantity database 16 stored in the reference feature quantity database 16 in association with coordinate values on the floor F, an imaging unit 11 for imaging a region on the floor F where the moving body is located at a predetermined time interval, A predetermined dot (center dot 5A) is selected from a plurality of dots 5 in the image acquired by the imaging unit 11, and positions of the selected dot 5A and a plurality of dots 5B located around the dot 5A The detected feature quantity calculating unit 17 that calculates the relationship as the detected feature quantity of the selected dot 5A, and the detected feature quantity are collated with the reference feature quantity data list 20 of the collation range of a predetermined size in the reference feature quantity database 16. By selecting the reference feature value having the highest degree of coincidence with the detected feature value, the feature value matching unit 18 is provided for detecting the absolute coordinate value of the selected reference feature value as the current position of the moving object on the floor F. It is configured. That is, in the present embodiment, since the method of encoding the dot pattern 3 corresponding to the coordinate value on the floor is not adopted, detection is performed using the pattern encoded corresponding to the coordinate value on the floor as in the past. Compared with the case where it performs, the freedom degree of the design of the pattern on a floor surface can be improved.

  Specifically, a random dot pattern can be arranged on the floor F as compared with the case where detection is performed using a pattern encoded in correspondence with coordinate values on the floor.

  In addition, since different dot patterns are placed on the entire floor surface in conventional position detection systems, it is necessary to lay the floor panels in order so that the order and orientation of the floor panels are correct when constructing the floor with materials such as multiple floor panels. However, in the moving body position detection system 1 of the present embodiment, the floor F is configured by arranging a plurality of floor panels f having the same dot pattern 3 in parallel. ing. By constructing as described above, since it is only necessary to pay attention to the direction of the floor panel f when constructing the floor F, it is possible to improve the workability when installing the floor panel f. Moreover, by using the floor panel f of the same dot pattern 3, a different dot pattern can be constructed at a lower cost than when each floor panel is applied. Furthermore, since the reference feature amount of the dot 5 of the entire floor F is stored in the reference feature amount database 16 in advance, it is not necessary to consider the boundary of the floor panel f.

  Furthermore, in the moving body position detection system 1 according to the present embodiment, the size (collation range) of the reference feature data list 20 to be collated with the detected feature is set to a range that does not include the dot 5 having the same reference feature. It is set. For this reason, when position detection is performed on the floor F where the same dot pattern 3 is repeated, the position of the moving body can be reliably detected without causing a situation such as erroneous detection or detection failure.

  In addition, in the moving body position detection system 1 according to the present embodiment, the size (collation range) of the reference feature quantity data list 20 to be collated with the detected feature quantity is set to the maximum moving speed of the moving body and the image acquisition of the imaging unit 11. It is set in a peripheral region that includes the movable range of the moving body determined from the time interval, and the collation range is set to the minimum necessary range. As a result, the calculation speed for detecting the position of the moving object can be increased.

  In the above-described embodiment, the example in which the mobile body position detection system of the present invention is applied to the position detection of a transport carriage or the like in a building has been described. However, the present invention is not limited to this, and a general plane The present invention can be widely applied to the position detection of the upper moving body.

It is a conceptual diagram of the moving body position detection system of this Embodiment. The left figure is a diagram showing an example of a dot pattern arranged on the floor surface, and the right figure is a diagram showing an example of an image obtained by imaging a partial region of the dot pattern. It is a top view of the floor constituted by arranging a plurality of floor panels side by side It is a conceptual diagram which shows the procedure of a mobile body position detection process. It is a figure explaining an example of the method of creating a dot pattern on a floor surface. It is a figure explaining an example of the method of creating a dot pattern on a floor surface. It is a figure explaining the standard feature-value of the dot in a dot pattern. It is a block diagram of the mobile body position detection system of this Embodiment. It is the figure which showed notionally the locus | trajectory of the moving body which moves on a floor surface. It is a flowchart which shows the flow of the process which the control part of a moving body position detection apparatus performs. It is a flowchart which shows the flow of the process which the control part of a moving body position detection apparatus performs. It is a flowchart which shows the flow of the process which the control part of a moving body position detection apparatus performs. It is a flowchart which shows the flow of the process which the control part of a moving body position detection apparatus performs.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Moving body position detection system 2 Floor surface 3 Dot pattern 4 Block 5 Dot 10 Moving body position detection apparatus 11 Imaging part 11a Illumination 12 Dot pattern creation part 13 Reference | standard feature-value calculation part 14 Reference | standard feature-value storage part 15 Control part 16 Reference | standard feature Quantity database 17 Detected feature quantity calculation unit 18 Feature quantity collation unit 19 Storage unit 20 Reference feature quantity data list f Floor panel F Floor

Claims (5)

A plane configured by arranging a plurality of panels in which the same dot pattern is arranged on the surface;
The positional relationship between the individual dots constituting the dot pattern and a plurality of dots positioned around the individual dots is calculated as a reference feature amount of the individual dots, and the reference feature amount is calculated as the individual dot. Reference feature amount storage means for storing in a reference feature amount database in association with coordinate values in the plane of
An imaging unit that is provided in a moving body that moves on the plane, and that images a region on the plane on which the moving body is located at a predetermined time interval;
A predetermined dot is selected from a plurality of dots in the image acquired by the imaging unit, and a positional relationship between the selected dot and a plurality of dots positioned around the dot is calculated as a detected feature amount of the selected dot. Detection feature amount calculating means for
By comparing the detected feature quantity with a reference feature quantity in a matching range of a predetermined size in the reference feature quantity database, the reference feature quantity having the highest degree of coincidence with the detected feature quantity is selected, and the selected reference Feature amount matching means for detecting a coordinate value of the feature amount in the plane as a current position of the moving object in the plane;
A moving body position detection system comprising: The dot pattern is
The moving body position detection system according to claim 1, wherein the panel is created by dividing the panel into a plurality of blocks and arranging dots at random positions in each block. The reference feature amount and the detected feature amount are:
An array composed of the direction of each line segment connecting a dot and a plurality of peripheral dots located in a circle area of a predetermined radius centered on the dot, and an array composed of the length of each line segment are included. Item 3. The moving object position detection system according to Item 1 or 2. The collation range is
4. The moving body position detection system according to claim 1, wherein the moving body position detection system is set in a range in which dots having the same reference feature amount are not included. The collation range is
4. The apparatus according to claim 1, wherein the moving object is set in a peripheral region including a movable range of the moving object determined from a maximum moving speed of the moving object and an image acquisition time interval of the imaging unit. The moving body position detection system described.

Images