Mobile body and mobile body position detection method

Abstract

To provide a mobile body that can be immediately operated through a little preparation in a new place and estimates a coordinate position with high accuracy.SOLUTION: An image processing unit 12 performs image detection by photographing, with sub-pixel accuracy with photographing devices 20a, 20b, ..., and 20n, a combination of one known marker 30 whose absolute coordinate position has been already acquired and which is registered by an operator and a plurality of unknown markers whose absolute position information is unknown but whose geometric structure is known. The absolute coordinate position of a mobile body 1 can be calculated through the detection of the known marker 30. At this point, the absolute coordinate position of the unknown marker is also calculated and the unknown marker is registered as a new known marker into a memory 14 together with the absolute coordinate position. The mobile body 1 can automatically increase the number of known markers while running and can perform self position estimation with high accuracy with reference to the large number of known markers 30 provided.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a moving body such as a robot or an automatic guided vehicle, and in particular to a detection technique for estimating a moving position with high accuracy. [Background technology]

[0002] Mobile objects such as robots and automated guided vehicles are increasingly being used in various environments, such as manufacturing sites, warehouses, public facilities such as buildings and airports, leisure facilities, local delivery, and construction sites. Such mobile objects are equipped with technology that moves autonomously to a destination, that is, autonomous driving control technology (autonomous movement technology). Autonomous driving control technology is broadly divided into self-location estimation technology and operation / driving planning technology, which collects (sensing) information about the mobile object itself and its surroundings, estimates the mobile object's own position and driving state, plans operations / driving in response to changes in the situation, and controls the driving mechanism. Among these, there are various methods for self-location estimation technology depending on the characteristics of the driving environment, available sensors, operating conditions, required accuracy, cost, and other constraints. For example, in terms of driving environments, the methods used differ depending on the environmental characteristics of outdoors, such as roadways, sidewalks, undeveloped spaces, next to buildings, and open sky, and even indoors, the methods used differ depending on the environmental characteristics of wide spaces, such as airports, factories, warehouses, corridors in buildings and hotels, and rooms in ordinary homes.

[0003] The sensors used include imaging devices (monocular, stereo, etc.) such as cameras that take still or video images, distance imaging devices that combine light-emitting devices and imaging devices, laser distance sensors (surveying instruments that accurately measure one point such as total stations, laser range sensors that obtain point clouds of the surrounding environment, etc.), millimeter wave radar, ultrasound, global positioning and measuring devices (GPS / GNSS) that sense the transmission and reception times of radio waves, inertial sensors (IMUs) that calculate the position and orientation using the acceleration of a moving object, and encoders that measure the amount of wheel rotation, and are selected according to the environmental characteristics, cost, accuracy, etc. For example, GPS can be used on roads with an open view above, but it cannot be used next to buildings or indoors.

[0004] Also, depending on the operating conditions, if an accurate map can be obtained in advance by the surveying instrument, etc., it is possible to use a map matching method, which estimates the position by comparing the map with real-time sensing information obtained by the moving body, but this cannot be used when the environment is prone to change, such as a construction site, and operation must be performed immediately in an unfamiliar location. In such cases where GPS and maps are not available, the SLAM method is used, which creates a map of the driving environment while estimating the position. In SLAM (Simultaneous Localization and Mapping), surrounding information is collected using a distance sensor or a camera, and feature points of shapes and images that serve as a reference for position estimation are detected, and the position is estimated by tracking the positions of the feature points while moving (odometry). Therefore, errors accumulate in the position estimated by SLAM and the map created, and the longer the distance or the wider the space, the greater the deviation from the true absolute coordinate position.

[0005] Therefore, it is necessary to correct the position using techniques such as combining position information obtained from multiple sensors, or passing through the same location again and performing map matching (loop closing) with a previously created map. However, there are limitations to the position accuracy, making it difficult to use this technique for aligning the installation position of equipment to the order of millimeters, for example at construction sites.

[0006] On the other hand, as a technique for performing position alignment with high accuracy, for example, Patent Document 1 discloses a method for correcting the position of a detection target and adjusting the attitude of a camera based on the difference between the detection position where the marker is captured and the reference position of a reference marker, although it is not for a moving body. Patent Document 2 discloses a method for estimating the relative position of a marker from a moving body. In Patent Document 2, the position and attitude of a marker are estimated by projecting the vertex positions of the marker on an image captured with an accuracy equal to or higher than the pixel accuracy of the imaging device (sub-pixel accuracy) onto a marker surface obtained from a distance sensor. [Prior art documents] [Patent documents]

[0007] [Patent Document 1] JP 2018-161726 A [Patent Document 2] JP 2015-56057 A Summary of the Invention [Problem to be solved by the invention]

[0008] As mentioned above, in a building construction site or the like, the installation position of equipment needs to be aligned to the millimeter order on a moving object. (a) There are cases where there is a ceiling and GPS cannot be used. (b) The environment is subject to change during construction, and the system must be operated immediately in an unfamiliar location, meaning that pre-prepared high-precision maps for position estimation cannot be used. (c) It is necessary to move within a wide area (100m x 100m) where pillars, materials, etc. are installed, and to estimate absolute position with millimeter-order accuracy. (d) There are constraints such as the inability to adopt expensive methods.

[0009] When using a surveying instrument using a laser, position accuracy can be obtained, but it is expensive, and problems arise such as being unable to work in areas where the laser is blocked by pillars, etc., and being able to measure the position but not the direction, so that construction line information cannot be obtained. In addition, when using the above-mentioned SLAM, errors accumulate due to the limit of the sensor resolution, position errors of feature points, and matching and tracking position errors of feature points, resulting in accuracy problems. Therefore, in this invention, a method using markers used for part alignment is applied and developed to estimate the position of a moving object. When a marker is used, the edge structure is known, so the marker position on the image can be estimated with an accuracy (subpixel) higher than that of the pixels of the imaging device.

[0010] The method disclosed in Patent Document 1, which is a prior art, photographs the markers at a close distance and from almost directly above, and in order to estimate the position and orientation of the moving body (self-position estimation), it is necessary to install a large number of markers at known positions on the floor surface in advance, which makes it impossible to solve the constraints (b) and (c) of the present invention. Furthermore, the method disclosed in Patent Document 2 requires a distance sensor in addition to an imaging device, so the accuracy and cost of the distance sensor are issues (constraints (c) and (d) of the present invention), and in particular, when estimating a marker surface several tens of meters away using a distance sensor, it is impossible to solve the constraint (c) of the present invention (absolute position accuracy on the order of millimeters). In addition, the disclosed method has a problem in that high sub-pixel accuracy cannot be obtained because it is necessary to use the vertices of the markers as feature points. Furthermore, in order to estimate the self-position of a moving body with high accuracy using this method, it is necessary to install a large number of markers at known positions in advance so that a marker with a substantially rectangular shape and a known absolute position can always be imaged at a position, for example, 2 meters away from the moving body, and it makes it impossible to solve the constraints (b) and (c) of the present invention.

[0011] The present invention has been made in consideration of the above-mentioned background, and has an object to realize a mobile body having a position and attitude estimation means that does not require a high-precision map for position estimation and can be immediately put into operation in an unfamiliar location with only minor preparation. Another object of the present invention is to realize a position detection method that can estimate the absolute position of a moving body with high accuracy while enabling movement over a wide range, and that enables position and orientation estimation at low cost by using markers, and to realize a moving body using the same. [Means for solving the problem]

[0012] In order to solve the above problem, the moving body of the present invention has one or more image acquisition means installed on the moving body and whose installation positions are known, a known marker whose edge structure information and absolute coordinate information in the space in which the moving body moves are known, marker structure information in which the geometric structure of the marker figure is expressed by a formula or the like and edge structure information are known but the absolute position coordinates of an unknown marker are unknown, and an edge detection means for calculating edge position information of the marker on the acquired image. The moving body calculates the position and orientation of the moving body using the absolute coordinate information and edge position information of the known marker included in the captured image in the moving body position and orientation calculation means. In addition, the marker position calculation means calculates the absolute coordinate position of the unknown marker from the moving body position and orientation calculated using the marker whose absolute coordinate position is known and the edge position information and marker structure information of the unknown marker by simultaneously or at once capturing the known marker and the unknown marker with the image acquisition means of the moving body. Furthermore, the control unit of the moving body registers the unknown marker whose absolute coordinate position is calculated by the marker position calculation means in the memory of the moving body as a known marker. The operator places a large number of unknown markers in or around the working area of ​​the moving object so that the unknown markers can be photographed together with the known markers by the image acquisition means. The marker image edge position calculation means calculates the position of the marker on the image with an accuracy of one pixel or higher by using edge structure information.

[0013] According to another feature of the present invention, when multiple known markers are included in images simultaneously captured by the image capture means, each known marker is compared with the current position of the moving body, and the absolute coordinate position of the moving body is calculated using the known marker that is closest to the moving body. Also, the multiple image capture means treats multiple images whose capture times are within a predetermined time difference as images captured simultaneously, and selects an image to be used depending on the movement situation. Only one or multiple known markers are installed to calculate the relationship between the absolute coordinate system in which the operator performs work on the moving body and the coordinate system of the center of the moving body when the moving body starts to move.

[0014] According to yet another feature of the present invention, structural information of known and unknown edges is processed in the moving body as a distribution of shading near the edge boundary or a distribution of its differential value. The unknown marker has an outline made up of a number of straight lines, and the marker structural information made up of a geometric structure can be expressed by an equation expressing at least the angle between the straight lines or the distance between the straight lines. In this way, the control unit of the moving body: a) Before starting movement, edge structure information and information on known markers whose absolute coordinate information is known and installed in the movement target area are acquired; b) acquiring images of known markers and / or unknown markers whose marker structure information and edge structure information are known but whose absolute coordinate information is unknown, using an imaging device, and calculating edge positions of the markers on the images; c) detecting known markers from the acquired image, and calculating the absolute coordinate position of the moving object based on the absolute coordinate information of the known markers and the edge positions; d) determining an absolute coordinate position of the unknown marker by calculating a relative position of the unknown marker with respect to the absolute coordinate position of the moving object from the marker structure information and the edge position; e) newly registering the determined unknown marker as a known marker in the control unit; f) The known markers registered in the control unit are used to repeat the processes in steps b) to e) above, thereby controlling the movement of the moving object. Effect of the Invention

[0015] According to the above-mentioned moving body of the present invention, a high-precision map for position estimation is not required for the operation of a robot or an automatic guided vehicle, and it is possible to immediately operate the robot in an unfamiliar place with only a little preparation. In addition, since the moving body can continuously perform absolute position estimation on the order of millimeters while moving over a wide range, it is possible to provide a moving body having a position and orientation estimation means with very few restrictions on the moving range and at low cost. The present invention is particularly effective in realizing the automation of work that requires the alignment of the installation position of equipment on the order of millimeters, particularly marking work, at building construction sites, etc. Marking work is work that marks the installation position of interior equipment, and conventionally, the position is specified by attaching thread with millimeter accuracy, and several hundred places per floor are marked. However, the work content of the moving body of the present invention is not limited, and the present invention is similarly effective in other work that requires alignment in warehouses, factories, etc., and work that requires precise estimation of the work position of the moving body. [Brief description of the drawings]

[0016] [Figure 1] 1 is a schematic configuration diagram of a position estimation system for a moving object 1 according to an embodiment of the present invention. [Diagram 2] 4 is a flowchart (part 1) showing the operations and processing procedure for performing position estimation using markers according to the present embodiment. [Diagram 3] 13 is a flowchart showing the operations and processing procedure for performing position estimation using markers according to the present embodiment (part 2). [Figure 4A] 1A and 1B are diagrams showing examples of the shapes of markers used in this embodiment, in which edges are formed by straight lines. [Figure 4B] 11A and 11B are diagrams showing examples of the shapes of markers used in this embodiment, the examples showing markers having curved edges. [Diagram 5]5A is a diagram for explaining, while comparing, methods for determining the structure and position of a part of an edge 311 of the marker 310 in Fig. 4. Fig. 5A is an enlarged view of an edge 311 obtained by cutting out a part of a linear portion on the left side of the square marker, Fig. 5B is a diagram showing brightnesses 302 and 312 of an image acquired by the imaging device 20, and Fig. 5C is a diagram showing a distribution of a differential value 305 obtained by differentiating the brightness 302. [Figure 6] 2 is a diagram for explaining the edge structure of the marker 30 of the present embodiment. FIG. [Figure 7A] 11 is a diagram for explaining geometric structure information of a marker 320 in this embodiment. FIG. [Figure 7B] 11 is a diagram for explaining geometric structure information of a marker 360 in this embodiment. FIG. [Figure 8A] 6 is a diagram for explaining an example of a method for calculating the absolute coordinate position of a moving object 1 using a known marker 604. FIG. [Figure 8B] 11 is a diagram for explaining an example of a method for calculating the absolute position coordinates of an unknown marker 610 using the absolute coordinate position of a moving object. FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0017] Hereinafter, a moving object according to an embodiment of the present invention will be described with reference to the drawings. In this specification, the front, rear, left and right directions will be described as directions shown in FIG.

[0018] FIG. 1 is an overall configuration diagram of a position estimation system for a moving body 1 according to an embodiment of the present invention. A position estimation device 10 is mounted on the moving body 1 such as a robot or an automatic guided vehicle. The moving body 1 has a drive source (not shown) and an energy source used when driving the drive source, and drives one of a plurality of wheels 3 to 6 by the drive source. For example, an electric motor is used as the drive source, and the right rear wheel 5 and the left rear wheel 6 are driven by independent motors (not shown) to enable not only forward and backward movement but also turning. The moving body 1 is provided with a drive mechanism (not shown) for the moving body 1 using a motor (not shown) and a movement control device (not shown) for controlling the drive mechanism, which may be known. When a motor is used, a power source (not shown) such as a battery is mounted on the moving body 1 and supplies power to the drive mechanism and the position estimation device 10 according to this embodiment.

[0019] The position estimation device 10 includes one or n (n is an integer greater than 1) imaging devices 20 (20a, 20b, . . . , 20n) and an information processing device 11. In the basic configuration of the present invention, highly accurate position estimation can be achieved using only one type of sensor, the imaging device 20. However, other sensors such as a sensor that measures the amount of rotation of any of the wheels 3 to 6, an acceleration sensor, an infrared distance sensor, a GPS sensor, etc. may be further provided, and the information processing device 11 may perform position and orientation estimation processing using the outputs of these sensors in combination.

[0020] The imaging device 20 is a means for acquiring an image, and is a known device that generates a still image or a continuous imaging signal from charges obtained from an imaging element such as a CCD image sensor or a CMOS image sensor, and is a still image camera or a video camera having an imaging range 21a with a predetermined angle of view. The imaging device 20 may be a monocular camera or a compound eye camera, and may be color or black and white. However, in order to detect the position of the marker 30 on the order of millimeters, it is necessary to select a device with high resolution while considering the balance with costs, and for example, a still image / monocular / black and white device is one option. For example, when a 4k camera (3,840 x 2,160 pixels) is used with a lens with a field angle of 50 degrees, one pixel at a position 5 m from the camera corresponds to 1 mm, and one pixel at a distance of 100 m corresponds to 2 cm. Therefore, in order to detect an object at a distance of 100 m with a resolution of 1 mm, a sub-pixel accuracy of 1 / 20 pixels is required. In this case, if two 4K cameras with lenses with a 25-degree angle of view are installed adjacent to each other with an angle of approximately 25 degrees between them, the field of view will be the same but the resolution will be doubled.

[0021] Moreover, if the range (distance from the moving body 1) of the object (marker) to be detected is limited to, for example, 50 m, the subpixel accuracy required for the imaging device 20 is halved. Therefore, as the imaging device 20, a camera with a wide angle of view may be installed for detecting the position of a nearby marker, and a camera with a narrow angle of view may be installed for detecting the position of a distant marker. Furthermore, as described later, since it is necessary to simultaneously detect a marker whose position is known and a marker whose position is unknown, it is advantageous for the imaging device 20 to have a wide field of view (front, back, left, right, etc.). In this way, the type, number, and layout of the cameras to be used are determined in consideration of the cost, camera resolution and angle of view (lens field of view range), subpixel processing accuracy, limitation of the marker detection range, marker placement policy, etc. At this time, data on the exact installation position of the camera relative to the reference coordinates of the moving body 1, its direction, and camera parameters (resolution, internal parameters: lens distortion correction value, focal length, etc.) are stored in advance in the memory 14 of the information processing device 11 as camera information, and are used for the position detection calculation process of the marker 30 by the image processing unit 12.

[0022] The information processing device 11 is, for example, a computer using a processor, and is configured to include an image processing unit 12 that processes images captured by the imaging device 20, an arithmetic and control unit (CPU) 13 that performs processing based on the image processing results, a memory 14, an input and display unit 15 such as a touch panel display, and a bus 16 that interconnects these components. The information processing device 11 performs image processing of image data captured by the imaging device 20 (20a to 20n), performs processing to estimate the self-position of the moving object 1, and outputs self-position information and operation commands to a movement control unit (not shown) of the moving object 1 by the image processing unit 12 and the arithmetic and control unit 13 executing a predetermined computer program.

[0023] Image data acquired by the imaging device 20 undergoes pre-processing such as distortion correction and brightness adjustment in the image processing unit 12 in response to a command signal from the calculation and control unit 13, and is temporarily stored as image information together with the imaging time, camera ID, etc. in the memory 14. The memory 14 is a collection of various storage media such as RAM, ROM, etc., including the main memory of the information processing device 11 and auxiliary memory devices such as storage.

[0024] In response to a command signal from the calculation / control unit 13, the image processing unit 12 performs various image processing such as edge detection processing in sub-pixel units from the image of the marker 30 and coordinate conversion processing based on the image information stored in the memory 14, pre-stored camera information of the imaging device 20, and information on the marker 30. At this time, intermediate images such as edge images are also stored in the memory 14.

[0025] The input / display unit 15 is, for example, a touch-type liquid crystal display, and is disposed so as to be visible and operable from outside the moving body 1. The input / display unit 15 displays the internal state of the moving body, the calculation status, the progress of the work, the work environment, the self-position, etc., and accepts input operations such as mode changes, various settings, and other command inputs by the operator.

[0026] The bus 16 can be configured with an IEBUS (Inter Equipment Bus), a LIN (Local Interconnect Network), a CAN (Controller Area Network), a LAN (Local Area Network), etc. In Fig. 1, the internal connection is expressed by a single line, but in actual implementation, it is bidirectional one-to-one communication, and each communication protocol may be different.

[0027] The mobile object 1 is further provided with a communication interface (not shown) and is capable of wireless communication with the outside. For the wireless communication, a known wireless communication standard such as wireless LAN or Bluetooth (registered trademark) may be used. The mobile object 1 is capable of two-way wireless communication with an external server 70 or a portable external operation and display terminal 80. The mobile object 1 directly accesses the server 70 to obtain information required for the traveling of the mobile object 1. The operation and display terminal 80 is composed of, for example, a portable tablet or smartphone, a portable or fixed PC, a network terminal device, etc., and the operation and display terminal 80 can access the server 70 and instruct the server 70 to transmit the information to the mobile object 1. In addition, by transferring information input from the input and display unit 15 of the mobile object 1 or information output to the input and display unit 15 to the operation and display terminal 80, it is also possible to check the input and output of the input and display unit 15 or to issue further input instructions from the operation and display terminal 80.

[0028] For example, when marking out work is performed using the mobile body 1 at a building construction site or the like, various data related to the construction, for example, any one of BIM (Building Information Modeling) data, CIM (Construction Information Modeling) data, and facility CAD (Computer Aided Design) data, is transferred from the server 70 to the operation and display terminal 80. If necessary, the operator uses the operation and display terminal 80 to create a movement / work plan for the mobile body 1 and inputs instructions to the mobile body 1. In this embodiment, at the start of work, the type of marker installed at the location where the work is to be performed, installation information, etc. can be transferred from the operation and display terminal 80 to the mobile body 1.

[0029] 2 and 3 are flowcharts showing the operations and processing steps for estimating the position of the moving body 1 in the moving body 1 of this embodiment. The processing steps up to calculating the initial position are shown in Fig. 2, and the subsequent processing steps up to continuously calculating the position after calculating the initial position are shown in Fig. 3.

[0030] First, the operator who operates the moving body 1 determines the absolute coordinates of the place (work area) where the work is to be performed using the moving body 1 (step 201). Here, the absolute coordinates represent the coordinate origin and the orientation of the x, y and z axes, and hereafter, the installation position of, for example, equipment is specified using the relative coordinate values ​​with respect to this coordinate origin and the orientation (attitude) with respect to the coordinate axes. Next, the operator installs a marker having marker information a representing the edge structure at a position where the absolute coordinates are known (step 202). Hereinafter, in this specification, a marker whose position information is known and whose absolute coordinates of the edge position are known will be referred to as "marker a" or known marker", and a marker whose absolute coordinate position is unknown will be referred to as "marker b" or unknown marker. When installing marker a for initial position calculation, the absolute coordinate position of the edge, which is the boundary line of the figure of marker a, is measured using a laser guide, distance measuring device, etc. The method for measuring the absolute coordinates of the first known marker for initial position calculation is arbitrary, and as another method, it may be set using a jig or a reference line used when determining the absolute coordinates (step 201). At least one marker a is set in the working area before the moving body 1 starts moving.

[0031] Next, the operator sets one or more markers b within the work area (step 203). Guidelines for setting marker b will be described later. Next, the moving body is set at an initial position for starting work (step 204). The setting position of marker a is a position where one or more of the mounted imaging devices 20 can capture an image of marker a from the moving body 1 facing the start direction of movement at the start point of movement. The method for recognizing this initial marker a may be specified by the operator inputting rough relative coordinates from the moving body 1, or the captured image displayed by the imaging device 20 of the moving body 1 may be displayed on the input / display unit 15 and the operator may specify marker a from the input / display unit 15 (for example, touching the part of marker a from the captured image displayed on a touch panel display device). In this state, when an initial position calculation command is sent to the calculation and control unit 13 of the moving body 1, the moving body 1 captures an image of the marker a (step 205), and the calculation and control unit 13 accurately calculates the initial absolute coordinate position of the moving body 1 and the attitude (direction of travel, inclination, etc.) of the moving body 1 by associating the edge position (in sub-pixel units) of the marker a on the captured image calculated using the camera information with the absolute coordinate position of the edge (step 206). The calculated initial position information of the moving body 1 is stored in the memory 14, and the initial position information is used when controlling the movement of the moving body 1 along a specified route.

[0032] Next, a process procedure for the moving body 1 to continuously calculate its position will be described with reference to the flowchart of FIG. 3. First, the moving body 1 starts moving in a pre-specified direction (step 211). At this time, the moving body 1 calculates its own coordinate position using images captured using markers a and b, moves along a pre-specified route, and performs a predetermined task at a destination position during the movement, such as marking out a black reference line on the floor. Next, the moving body 1 captures an image of marker a when stopped or while moving (step 212), and calculates the latest coordinate position and orientation of the moving body 1 (step 213). It is preferable to perform steps 212 and 213 substantially continuously at short intervals during the movement. At this time, since there is a risk of blurring of the image when the image is captured while moving, the slower the moving speed of the moving body 1, the better the position calculation accuracy. Therefore, when accuracy is particularly required for calculating the position and orientation, it is better to capture the image while stopped. The imaging of the marker a (step 212) and the calculation of the position of the moving object 1 (step 213) are the same processes as those in steps 205 and 206 described above with reference to FIG. 2, which were first performed.

[0033] When the marker a is captured, the calculation and control unit 13 must identify which captured marker corresponds to the specific marker a registered in advance in the moving body 1. First, if the marker a captured in the past and already identified can be tracked to where it is in the current image, this tracking information is used. The tracking information is, for example, a history that records the position of the marker a in the captured past image. If the tracking information is not registered, the information of the registered marker a and another marker b captured in the captured image is matched to the information of the other marker b based on the approximate position and orientation (position and direction) of the current moving body 1 and the absolute coordinates of the marker a. If the marker b is not captured at the same time as the marker a is captured (No in step 214), the processing of steps 212 and 212 is repeated to store the trajectory of the position of the moving body 1 in the memory 14. At this time, the calculation and control unit 13 of the moving body 1 determines whether the current position and operation of the moving body 1 are appropriate for the route or destination set in advance or instructed by the operation and display terminal 80 or the like, and controls the traveling of the moving body 1.

[0034] If marker a and marker b are simultaneously imaged in step 214 (Yes in step 214), the coordinate position of marker b is calculated from the position and orientation (six dimensions of position and orientation) of the moving object 1 calculated using marker a, marker information b consisting of the edge structure and geometric structure of marker b, and the edge position of marker b on the captured image (in sub-pixel units) (step 215). The method of calculating the position of marker b will be described later. Then, when the absolute position of the newly imaged marker b can be calculated, the marker information b of marker b and its position are registered in memory 14. At this time, since the absolute coordinate position of this marker b is now known, it is re-registered as marker a (step 216).

[0035] After step 216 is executed, the number of markers a registered in the memory 14 and identifiable by the calculation / control unit 13 is increased by one. If multiple markers b are detected in the captured image in step 214, the processes of steps 215 and 216 are executed for the number of detected markers b, thereby registering new markers a in the memory 14. When the execution of step 216 is completed, the process returns to step 211. The control of steps 211 to 216 is continued until the operation of the moving body 1 is completed. Since the movement control (step 211) of the moving body 1, such as forward and backward movement, is continuously performed in parallel with the process for calculating the position in FIG. 2 and FIG. 3, when the process of steps 211 to 216 is completed and the process returns to step 211, the moving body 1 has often moved to a new position.

[0036] In order to calculate the position of the moving body 1, it is sufficient that at least one marker a is photographed in the captured image acquired from any one of the imaging devices 20a to 20n. In this case, if many markers a can be photographed, the reliability of the accuracy of the calculation of the position of the moving body 1 is increased. However, since the accuracy of the calculation of the position of the moving body 1 is higher when the marker a closer to the moving body 1 is used, the distances between the multiple markers a and the moving body 1 may be compared, and one or more of the closest markers a may be used for the position calculation. It is possible that the marker a cannot be photographed at all during the movement of the moving body 1, and during that time, the accurate position of the moving body 1 cannot be calculated. Even if the position cannot be calculated (short), the moving body 1 continues autonomous traveling by the movement control device (not shown), and when the marker a can be photographed again, the calculation and control unit 13 calculates the accurate position of the moving body 1 again.

[0037] In calculating the position of marker b, the result of calculating the position of the moving object 1 by marker a is used, so that marker b and marker a need to be photographed at the same time. That is, before it becomes necessary to calculate the position of the moving object 1 using marker b, marker b needs to be arranged so as to appear in the same captured image as marker a, or the moving path of the moving object 1 needs to be generated so that marker b appears in the same captured image. In addition, the captured image is not limited to that obtained from one imaging device 20, and a collection of images obtained from multiple imaging devices 20a to 20n can be used as one captured image. Therefore, it is desirable that the imaging times of the multiple imaging devices 20a to 20n used as one captured screen are synchronized. Even if they are not completely synchronized, it is sufficient to be able to select images acquired within a predetermined time difference. In this case, in consideration of the processing load, even if all the imaging devices 20a to 20n are not synchronized, multiple sets of synchronized imaging devices may be prepared, or only images that are synchronized (acquired within a predetermined time difference) may be used to create the one captured image. Furthermore, if the moving object 1 is stationary, images from all of the imaging devices 20a to 20n can be used even if they are not synchronized, so that images to be used can be arbitrarily selected according to the moving state.

[0038] An example of a guideline for arranging the marker b is to arrange a plurality of markers b so that the next marker b can be photographed before the current marker a cannot be photographed on the moving path of the moving body 1, taking into consideration the installation status of the imaging device 20 of the moving body 1, i.e., the range of the captured image, the distance from the moving body to detect the markers a and b, etc. In addition, the positions of the markers a and b placed by the operator may be roughly set in advance or appropriately by the operation / display terminal 80, the shapes of the markers a and b at the start of the movement may be different, and the moving body 1 may identify whether the marker is a, b, or a marker a that has been changed and reregistered from marker b based on the shapes of the markers a and b, or the markers may be assigned a recognizable ID (image pattern, color pattern, character, symbol, etc.) determined in advance and registered in the memory 14. In addition, the markers a and b may be moved and rearranged while the moving body 1 is moving. According to the above method, only one marker a, which is detected first, needs to have its location information prepared in advance, and the other marker b only needs to be placed or removed according to the route of the moving body 1, making it easier for the operator to make advance preparations when performing work using the moving body 1.

[0039] FIG. 4 is a diagram for explaining an example of the shape of the marker used in this embodiment. FIG. 4A shows a group of markers formed by combining straight lines, and all the markers 30, 310 to 350 are of any color (for example, black) and are formed as planar, visible figures having multiple straight line portions. The multiple straight line portions included in the outer edge (boundary) of the markers 30, 310 to 350 form corners at the intersections of the straight line portions. It is preferable that the markers 30, 310 to 350 are fixed so that their shapes do not change when detecting the position of the moving body 1 during its movement. For example, a plate with a flat surface may be fixed to the portable member, and the markers 30, 310 to 350 may be formed on the plate with paint. Also, the markers 30, 310 to 350 formed of a sealant may be attached to the plate. The method of fixing the markers 30, 310 to 350 is arbitrary, and the markers 30, 310 to 350 made of a sealant may be affixed to objects (for example, pillars or walls) within the movement area in which the moving body 1 moves.

[0040] In FIG. 4A, the marker 30 is a right triangle having three straight line segments, and the marker 310 is a square having four straight line segments. The markers 320 and 330 are parallelograms having four straight line segments, and the marker 330 is a shape called a rhombus in which the lengths of two adjacent sides are equal. The marker 340 is a regular pentagon having five straight line segments, and the marker 350 is a regular hexagon having six straight line segments. In FIG. 4A, the outer edge portion of each of the markers 30 and 310 to 350 is formed with an edge shape having a gradation or an edge shape having no gradation. Although the color of the markers 30 and 310 to 350 is not clear from the description of FIG. 4A, they may be formed with any color. However, it is preferable that the color is such that the image processing is easy in the moving object 1. The shape of the markers 30 and 310 to 350 is any shape as long as it has a plurality of straight line segments on the outer edge portion, and the marker may have a shape other than the illustrated shape.

[0041] FIG. 4B shows an example of a group of markers formed with a curved portion on the outer edge. Marker 360 is an ellipse. The shape of the edge of marker 360 does not include a straight line, but if it has a known edge defined by the structural formula of a conic section or the like, it can be used as a marker even if it does not include a straight line portion. However, it is desirable for the marker to have a straight line portion in order to improve the detection accuracy of the edge of the marker using a captured image and to facilitate the calculation for detection (a straight line is a straight line when viewed from any direction). Marker 370 has a shape like a cut-out part of an ellipse, and is configured with one curved portion and one straight portion that can be defined by a conic section. Here, an ellipse is shown as an example, but in addition to ellipses, curves expressed by conic sections include circles, parabolas, and hyperbolas, and such curves can also be used in the structural formula of a conic section. The shape of markers 360 and 370 is arbitrary as long as it has a known edge defined by a structural formula, a graphic template, or the like on the outer edge, and markers of other shapes than those shown in the examples may be used.

[0042] FIG. 5 is a diagram for explaining the edge structure of the marker 310 (see FIG. 4A). FIG. 5A is an enlarged view of an edge 311 obtained by cutting out a part of the left linear portion of the square marker. The right side shows a part of the edge 311 of the monochromatic marker 310, and is an example in which the boundary portion of the marker 310, i.e., the outline portion of the edge 311, changes sharply in shade, and is shown in two gradations, 1 and 0. For comparison, the left side shows an edge 301 with a gradation in which the change in shade is gentle. The edge 301 is a part of the marker whose boundary portion is displayed using pixel gradations. The markers having these edges 301 and 311 are installed in positions that can be seen from the moving range of the moving object 1, i.e., positions that can be photographed by at least one of the imaging devices 20a to 20n.

[0043] FIG. 5B is a diagram showing the brightness 302 and 312 of the image acquired by the imaging device 20. For the sake of simplicity, in FIG. 5B and FIG. 5C, the edges 301 and 311 are shown as images 303 and 313, which are three pixels in the vertical direction and five pixels in the horizontal direction. As illustrated in FIG. 4, the structure of the edges 301 and 311 of the marker may include information on the edge shading distribution as shown in the edge 301, in addition to macro shape information such as a straight line or a known curve. For example, the edge 301 (enlarged view) of the marker whose contour portion has a continuously changing shade distribution is drawn with a gentle brightness 302. The horizontal axis of the graph of the brightness 302 corresponds to the horizontal position of the edge 301, and the vertical axis corresponds to the brightness of the edge 301. The image 303 of the edge 301 acquired by the imaging device 20 is digital data of 3×5 pixels having shade information corresponding to the brightness 302. The diagram below the image 303 shows, in a bar graph, the brightness 304 acquired as the image 303 from the imaging device 20 against the brightness 302. The captured image 303 has the same information for three rows in the vertical direction, so the diagram shows only the brightness for one row. The brightness 304 of each pixel of the image 303 acquired by the imaging device 20 changes in stages, as can be seen from the size of the bar graph.

[0044] 5C is a diagram showing the distribution of differential values ​​305 obtained by differentiating luminance 302. Differential values ​​305 are obtained by differentiating the curve of luminance 302 of edge 301. The value of differential value 305 is zero where there is no change in luminance 302, and the value increases in the negative direction when the slope of the curve of luminance 302 to the lower right is large. Image 306 is calculated by differentiating image 303 obtained by imaging device 20, and shows the differential value of each pixel as a gray value. The graph below image 306 shows differential values ​​307 of this image 306 as a bar graph.

[0045] The diagrams in the right column of FIG. 5B show the characteristics of a marker 310 (see FIG. 4) having an edge 311 with a steep gray-scale change. At the edge 311, the gray-scale distribution 312 is outlined by a steep line, and in an image 313 obtained by the imaging device 20 at this time, the pixels in the boundary portion are intermediate values ​​indicated by light gray-scale, while the rest of the image is either high or low brightness. Parts with high brightness are photographed as white pixels, and parts with low brightness are photographed as black pixels, resulting in an image in which the pixels in the boundary portion have an intermediate density. In an image 316 showing the differential value of the image 313, only the pixels including the boundary are intermediate values, while the rest are constant values ​​(gray-scale values ​​indicating zero). The brightness of this image 316 is shown in a bar graph in the graph below the image 316.

[0046] In FIG. 5C, the positions of the edges obtained from the images 303 and 313 are shown as edge positions 300a to 300d. In this case, in order to obtain the edge positions in sub-pixels, the luminance distribution or differential value distribution of the edge is held as a template, and the template is fitted to the luminance or differential value of each pixel to calculate the edge position. For example, in the case of luminance 304, the position with the smallest difference is searched for while shifting the position of the luminance distribution template (corresponding to luminance 302), and the edge position is obtained from the fitted position. This is similar processing to other cases (luminance 314, differential values ​​307, 317, 318). In this case, the most accurate sub-pixel calculation is case 307, in which the edge with a gentle change is differentiated. This is because there are extreme values ​​that change continuously. The least accurate are cases 317 and 318, in which the edge with a steep change is differentiated. In cases 317 and 318, the differential values ​​of each pixel are the same, but the calculated edge positions are different. Thus, the sub-pixel calculation value of the edge position obtained by the method in this case has a large error. Also, in case 314 where the brightness of the steep edge is used, the sub-pixel value depends on the brightness value of the pixel where the edge exists, so the accuracy is affected by noise.

[0047] The case of brightness 304, which uses the brightness value of the edge with a gradual change, has the second highest accuracy after the case of brightness 307, because the brightness distribution is fitted over several pixels. In this way, the position of the edge can be obtained in sub-pixels using a template. Here, an example of a vertical straight line is shown for simplicity, but in reality, as in the marker 30 in FIG. 6, the edge 31 (image in the vicinity of the edge position 30e) is obtained as an image that crosses the pixel in various directions and positions. Therefore, the fitting point is searched for for the pixel of the edge image 31 while changing the position, direction, and stretching of the sub-pixel calculation template. In addition, for the sub-pixel fitting, methods such as SSD parabolic fitting that minimizes the square error and POC phase-only correlation using Fourier transform can also be used. For the shape of the edges 31, 301, etc., a straight line is preferable because of the accuracy and ease of calculation (a straight line is a straight line when viewed from any direction), but a curved line can also be used as long as it can be registered as a template. In this way, by having edge luminance distribution, differential value distribution, templates, etc. as edge structure information of the marker, the edge position of the marker can be found with sub-pixel accuracy.

[0048] Sub-pixel position accuracy information can be expressed as an error value during fitting. Therefore, this accuracy information can be used to select and use locations with high accuracy (calculated values). It is also possible to use time-series information of the image as a method for improving accuracy. For example, by tracking the position that changes over time after fitting once, the effect of noise in each pixel can be reduced and the calculation load can also be reduced. In addition, there are cases where the moment when an edge crosses over between pixels can be identified using time-series information, and accuracy can be improved by using edge detection information of pixel boundary positions (sub-pixel accuracy). Furthermore, accuracy can be improved by using long areas such as straight lines, or by integrating and fitting several edges with different directions.

[0049] In order to calculate the positions of the markers 30, 310 to 370, etc., it is necessary to have geometric structure information that uniquely defines the structure of the marker and indicates the relative positional relationship of the edges of the marker. The geometric structure information includes, for example, information on the angle and length of a line. The geometric structure information may be represented in any manner as long as the structure can be uniquely defined, and an example is shown in FIG. 7A.

[0050] Fig. 7A is an example of a parallelogram marker 320 (see Fig. 4A). When detecting edges with sub-pixel accuracy, it is possible to detect a point on an edge (e.g., Pa1, Pa2) or a vector of an edge (e.g., vector a) on an image. The relational equation of the edges is then used as the geometric structure information. Edges 501 and 502 are in an angle θ relationship, so the structure is given by Equation 1 using the dot product of the vectors.

[0051]

number

[0052] Similarly, the relationship between edge 501 and edge 503 has the structure of formula 2.

[0053]

number

[0054] As for length information, the distance between side 501 and side 503 is defined as L, and the vector from an arbitrary point Pa1 on side 501 to an arbitrary point Pc2 on side 503 is defined as vector c. Using the cross product of the vectors, the structure of formula 3 is obtained.

[0055]

number

[0056] Fig. 7B is an example of an elliptical marker 360 (see Fig. 4B). As a structural formula of a conic section showing the ellipse in Fig. 7B, for example, Formula 4 can be used.

[0057]

number

[0058] Here, r is the distance between the focal point F 510 and a point on the edge (for example, P1, P2, P3), L is a value called a half chord (the Y coordinate value at a point on the edge whose element is the same value as the X coordinate of the focal point F), e is a value called eccentricity (the distance between any point on the edge and the focal point: the distance between the point on that edge and the directrix 511 = e:1), and θ is the angle between the point on the edge and the X axis, with the focal point 510 as the center.

[0059] Even if the marker has a part of the shape of the parabola 361 (e=1) or hyperbola 362 (e>1) shown in Figure 7B (for example, a shape similar to marker 370 in Figure 4B), the position of the marker can be calculated by using the structural formula of a conic section in a similar manner using the above-mentioned formula 4.

[0060] 8A and 8B are diagrams for explaining the position calculation method of the present invention. Fig. 8A shows an example of calculating the position and orientation of a moving body 1 using a marker 604 whose absolute coordinate position is known (marker a), and Fig. 8B shows an example of calculating the absolute position and orientation of an unknown marker 610 (marker b) when the position and orientation of the moving body 1 is known.

[0061] First, an example of calculating the position and orientation of the moving object 1 using a marker 604 whose absolute coordinate position is known will be described. In calculating the position of the moving object 1, since the absolute position of the known marker 604 is known, there is no need for geometric structure information of the known marker 604. Since the installation position of the image capture device 20 mounted on the moving object is accurately known in advance, the relationship between the camera-centered coordinate system and the coordinate system of the moving object 1 is stored in advance as camera information. Therefore, it is sufficient to calculate the position and orientation in the camera-centered absolute coordinate system here.

[0062] In FIG. 8A, for a camera coordinate system 602 at the camera center C601, an image 603 is called a normalized image, and is an image whose center coordinates are (0,0,1) and whose coordinates are expressed in XY plane coordinates (units are distance, not pixels). This coordinate system is transformed from the coordinate system of the optical element, which is the raw data of the imaging device, using internal parameters such as focal length. An edge 605 on one side of a known marker 604 is imaged on the image 603, and an arbitrary position (P1, P2) in sub-pixels is detected. Here, it is not known which point on the edge 605 is imaged as the arbitrary position (P1, P2) in sub-pixels. Since any sub-pixel position (P1, P2) is within the same plane passing through the edge 605 and the camera center C, equation 5 is established using its normal vector n607 (the normal n607 of the triangle with vertices C, P1, and P2 is perpendicular to the line connecting any point M1 on the edge 605 whose absolute coordinate position is known and the camera center C).

[0063]

number

[0064] This formula is composed of three formulas for x, y, and z elements, but contains six unknowns, namely, the rotation matrix R (roll, pitch, yaw) and the translation matrix T (x, y, z), as coordinate transformations expressing the position and orientation of the moving object 1. Therefore, it is necessary to obtain R and T by utilizing the parameters (Px, Mx) of the other sides 606 in a similar manner. However, in the above-mentioned example of the construction site, the roll, pitch, and height x are almost constant values. Therefore, the marker a only needs to have three straight lines (not necessarily a figure). In this example, the normal, which is a common element between the edge position on the image and the edge position of the marker, is used, but other common elements according to the edge shape may be used (for example, the normal direction always passes through the center of a circle). This method does not require matching of the edge position (point) on the image and the edge position (point) of the marker, as in the SLAM technology, and therefore it is possible to calculate the position of the moving object 1 with high accuracy without generating errors in detecting specific points such as vertices and corners, and errors caused by matching points detected on the marker and the image.

[0065] Next, an example of calculating the absolute position and orientation of an unknown marker 610 (marker b) when the position and orientation of a moving object are known will be described. In Fig. 8B, the relationship between the camera center C 601, the camera coordinate system 602, and the captured image 603 is the same as in Fig. 8A. Here, it is possible to use Relational Expression 6 for projecting the edge position P3 detected on the image 603 onto the absolute coordinate system (point M3 on the edge of the unknown marker exists on the straight line passing through the camera center C and the edge position P3).

[0066]

number

[0067] This equation is composed of three expressions of x, y, and z elements, but the position and orientation of the moving object 1, i.e., the rotation matrix R (roll, pitch, yaw) and translation matrix T (x, y, z) of the coordinate transformation, and the point Px on the edge on the image (vector CPx: the positions on the image of P3 and P4 in FIG. 8B) are known, and the value k obtained by dividing the distance from the camera center to the point Mx on the edge of the unknown marker by the distance from the camera center to the edge position Px on the image, and the absolute coordinates (x, y, z) of the edge position of the marker are unknown. Therefore, the geometric structure information of the marker is required. The absolute coordinate position of the marker can be specified by associating multiple edge positions Px (exemplified as P3 and P4 in FIG. 8B) with the edge position Mx of the unknown marker 610 (exemplified as M3 and M4 in FIG. 8B) and substituting multiple edge positions Mx on the unknown marker into the geometric structure information of the marker, such as expressions 1 to 4. This method also does not require matching the edge positions (points) on the image with the edge positions (points) of the marker, as is done with SLAM technology, so there are no errors in detecting specific points such as vertices and corners, and no errors caused by matching points detected on the marker and the image, making it possible to calculate the position of the moving object 1 with high accuracy.

[0068] According to the above-mentioned moving body of the present invention, in the operation of a robot or an automatic guided vehicle, the moving body 1 moves within a working area while repeating the process of calculating the absolute coordinate position of an unknown marker whose absolute coordinate information is unknown from an initially set known marker and newly registering the unknown marker as a known marker in a memory unit. Therefore, the operator does not need to set a plurality of known markers in the working area in advance, and only needs to set a large number of unknown markers. In addition, according to this embodiment, a GPS and a high-precision map for position estimation are not required, so that the moving body can be operated immediately in an unfamiliar place with only a little preparation. Furthermore, it has become possible to realize a moving body having a position and orientation estimation means capable of performing absolute position estimation on the order of millimeters while being movable over a wide range. [Explanation of symbols]

[0069] 1. Mobile 2. Body 3~6 wheels 10 Position estimation device 11 Information processing equipment 12 Image processing section 13 Calculation and control unit (CPU) 14. Memory 15 Input / display section 16 Bus 20, 20a to 20n Imaging device 30 Marker 31 Edge Images 70 Servers 80 Operation and display terminal 300 Markers 300a~300d Edge position 301 Edge 302, 304 Brightness 303, 306 images 305, 307 Differential value 310 Marker 311 Edge 312, 314 Brightness 313, 316 images 315, 317, 318 Differential values 320~370 marker 361, 362 hyperbola Area 501~503 510 focus 511 directrix 601 Camera center C 602 Camera Coordinate System 603 images 604, 610 Marker 605 Edge 606 sides

Claims

1. A moving object that performs work while measuring its own position, one or more image acquisition means that are installed on the moving body and capture images of known markers, whose edge structure information and absolute coordinate information, which is the position of the marker in the working space, are known, and unknown markers, whose marker structure information and edge structure information, which indicate the shape of the marker, are known but whose absolute coordinate information is unknown; An edge detection means for calculating position information of an edge on a photographed image of the marker; a position and orientation calculation means for calculating a position and orientation of the moving object using absolute coordinate information of the photographed known markers and position information of the edges; a marker position calculation means for calculating an absolute coordinate position of the unknown marker from a position and orientation of the moving body calculated using the known marker, edge position information of the unknown marker, and the marker structure information by photographing the unknown marker and the known marker at the same time, Calculating position information of the edges of the known markers and the unknown markers with a pixel accuracy of the image acquisition means or higher using the edge structure information; the marker position calculation means registers the unknown marker, whose absolute coordinate position has been calculated, in an information processing unit of the moving body as an additional known marker; a position and orientation calculation means for measuring the position and orientation of the moving body using a known marker added by registration, and a marker position calculation means for calculating an absolute coordinate position of the unknown marker, even if the first known marker is no longer captured by the image acquisition means due to the movement of the moving body.

2. The moving body according to claim 1 , When a plurality of known markers are included in the images simultaneously captured by the image acquisition means, the present position of the moving body is compared with each of the known markers, and the absolute coordinate position of the moving body is calculated using the known marker that is closest to the moving body.

3. The moving body according to claim 1 , A moving body characterized in that the multiple image acquisition means treats multiple images whose acquisition times are within a predetermined time difference as images taken simultaneously when the moving body is moving, and treats multiple images acquired after the moving body has stopped as images taken simultaneously regardless of the acquisition time difference when the moving body is stopped, and selects the images to be used depending on the moving status.

4. The moving body according to claim 1 , A moving body characterized in that one or more known markers are installed at the start of operation of the moving body for setting an initial position to calculate the relationship between an absolute coordinate system in which the moving body performs its work and a coordinate system at the center of the moving body.

5. The moving body according to claim 1 , The moving object, wherein the edge structure information is a grayscale distribution near the edge boundary or a distribution of its differential values.

6. The moving body according to claim 1 , A moving body, wherein the unknown marker has a contour consisting of a plurality of straight lines, and the marker structure information consisting of a geometric structure is an equation that represents at least the angle between the straight lines or the distance between the straight lines.

7. The moving body according to claim 1 , A moving body, characterized in that the unknown marker is placed at a position on the moving path of the moving body where a situation in which it can be imaged simultaneously with one or more of the known markers can be maintained.

8. A moving body having an imaging device for acquiring image information and a control unit for detecting a marker in a captured image to detect an absolute coordinate position of the moving body, The control unit is a) acquiring edge structure information and information of known markers whose absolute coordinate information is known and installed in the area to be moved before the movement starts; b) acquiring images of the known markers and / or unknown markers whose marker structure information and edge structure information are known but whose absolute coordinate information is unknown by the imaging device, and calculating edge positions of the known markers and / or unknown markers from the acquired images; c) detecting a known marker from the acquired image, and calculating an absolute coordinate position of the moving object based on the absolute coordinate information of the known marker and the edge position; d) determining an absolute coordinate position of the unknown marker by calculating a relative position of the unknown marker with respect to an absolute coordinate position of the moving body from the marker structure information and the edge position; e) newly registering the determined unknown marker as a known marker in the control unit; f) A method for detecting a position of a moving body, comprising repeating the processing of steps b) to e) by using the known marker registered in the control unit to control the movement of the moving body.

9. 9. A position detection method according to claim 8, comprising: In the step d), the control unit captures images of the known markers and the unknown markers at the same time with the imaging device of the moving body, and calculates an absolute coordinate position of the unknown marker from the position and orientation of the moving body calculated using the known markers, and edge position information and marker structure information on the image of the unknown marker.

10. 9. A position detection method according to claim 8, comprising: In the step c), when a plurality of known markers are imaged by the imaging device, the distance between each of the known markers and the moving body is compared, and the absolute coordinate position of the moving body is calculated using the nearest known marker.

11. 9. A position detection method according to claim 8, comprising: A plurality of the imaging devices are provided, A position detection method for a moving body, characterized in that in step b), when the moving body is moving, the control unit treats multiple images whose acquisition times are within a predetermined time difference as images taken simultaneously, and when the moving body is stopped, treats multiple images acquired after the moving body has stopped as images taken simultaneously regardless of the acquisition time difference, and selects the images to be used depending on the movement status.

12. 9. A position detection method according to claim 8, comprising: When the moving body starts to move, one or more of the known markers are provided for setting an initial position for calculating a relationship between an absolute coordinate system for performing an operation of the moving body and a coordinate system of a center of the moving body; A method for detecting a position of a moving object, wherein the unknown marker is formed to include a plurality of straight lines, and the marker structure information is an equation that represents at least the angle between the straight lines or the distance between the straight lines.

Images