JP4448497B2 - Self-position detecting device and position detecting system for moving body - Google Patents

Description

  The present invention relates to a self-position detection device and a position detection system for a moving body.

  Conventionally, a moving body such as a self-propelled vehicle or a mobile robot (hereinafter simply referred to as a robot) detects a position of the mobile body by processing an image captured by a camera provided on the mobile body. Detection systems are known (see, for example, Patent Document 1, Patent Document 2, and Non-Patent Document 1).

  The position detection device described in Patent Document 1 is provided in, for example, a self-propelled industrial robot used in a factory. And in order to detect specific positions, such as a corner in a factory, only the fluorescent lamp just above a specific position is previously covered with the polarizing filter among the illuminations (fluorescent lamps) installed in the ceiling. This position detection device includes a camera for photographing the ceiling, and detects that a specific position has been reached by imaging a fluorescent lamp covered with a polarization filter and performing image processing. As a result, an industrial robot equipped with a position detection device can draw a movement trajectory that, for example, usually moves straight along a line, but changes straight ahead again when detecting a specific position. It becomes possible.

  The position detection device described in Patent Literature 2 is provided in an autonomous mobile robot that can autonomously move according to a predetermined command, for example, at home as well as in a factory. This robot moves by itself to a battery charging stand (stand) provided in the movement area in accordance with a command. In order to perform charging, it is necessary to accurately connect the battery supply plug installed in the stand and the battery reception plug provided in the robot. However, the robot alone cannot accurately move to a desired position. Therefore, this stand has three spherical markers at a position higher than the height of the robot. On the other hand, the robot is provided with a camera for upward shooting at the top of the head, and an accurate position alignment with the battery supply plug is possible by imaging a marker with the camera and processing the image.

  The position detection device described in Non-Patent Document 1 is provided, for example, in an industrial robot capable of self-running. As an example, a plurality of reflective seals are pasted in advance on a ceiling above a desired place (referred to as a change point) where the movement trajectory is to be changed in the factory. An ID for identification is given to the reflection seal, and information such as a command is stored. The position detection device includes a light source that irradiates the ceiling with visible light, a camera that captures the reflection sticker that reflects the irradiation light, and a comparison table of the ID and command of the reflection sticker, and processes the captured image. By doing so, the ID of the reflective seal is specified, and the command is extracted from the comparison table. Therefore, the robot can change its operation each time it reaches the change point and obtains a command from the reflective sticker. For example, by preparing commands indicating straight travel, rotation, stop, etc. for each reflective sticker and combining them appropriately, the robot can perform complex motions and smooth motions.

  In addition, in a wearable computer (wearable PC) that a human (user) can wear on the lower back, for example, instead of a robot, an image captured by a camera with a light source mounted on the user's head is processed by the connected wearable PC. By doing so, a navigation system in which the user can detect his / her position is known (for example, see Non-Patent Document 2).

In the navigation system described in Non-Patent Document 2, for example, a plurality of position specifying markers are regularly pasted regularly on the ceiling of the user's moving area. For example, markers of about 15 cm on a side are arranged in a plover shape at intervals of about 50 cm. This marker is obtained by pasting a retroreflective member sufficiently smaller than the base in a predetermined pattern on a translucent base (substrate). For example, it is possible to assign, for example, 4096 IDs to the marker by arranging one reflecting member or a maximum of 4 × 4 grids on the base of the marker. The wearable PC specifies the marker ID by processing the captured image. Therefore, by associating the ID and the position information, the user can recognize the current position every time the ID is acquired from the marker.
JP-A-8-300309 (paragraphs 0011 to 0016, FIG. 1) JP 2004-303137 A (paragraph 0044 to paragraph 0052, FIG. 1) Yuji Mezaki, 1 other, "Navigation using reflective stickers for mobile robots", IEICE, Computer Vision, January 24, 1992, 76-21, p.151-158 Yusuke Nakazato, 2 others, “Identification of user's position and orientation using retroreflective marker and infrared camera”, IEICE Technical Report, July 2004, IE2004-24, p.25-28

  However, since the position detection device described in Patent Document 1 detects the position of an illumination (fluorescent lamp) provided at a specific position, it is effective for line tracing of an industrial robot or the like. When applied to the robot, the behavior of the robot is restricted. Further, if an object that actively emits light is disposed instead of the fluorescent lamp, the cost of the entire system increases.

  In addition, the position detection device described in Patent Document 2 enables accurate position detection by a robot that can move autonomously, but in order to detect a plurality of places in a moving region, a plurality of positions corresponding to the places are detected. Marker (or stand) is required, and the entire system is expensive.

The position detection device described in Non-Patent Document 1 detects a command at a change point and is effective for an industrial robot whose movement trajectory is determined in advance, but is applied to an autonomous mobile robot. In such a case, the behavior of the robot is restricted. Note that the reflective mark requires a storage area in which commands can be stored, which increases the cost of the reflective mark.
Further, the navigation system described in Non-Patent Document 2 has a problem that the error between the marker detection position and the actual marker arrangement is several tens of centimeters or more, and the error is large.
Furthermore, in a conventional position detection system that needs to place a member (marker) for specifying a position on the ceiling or the like, there is a demand for reducing the effort required to place the marker in order to generalize the position detection system. .

Therefore, an object of the present invention is to solve the above-described problems and provide a self-position detection device and a position detection system for a moving body that can accurately detect an arbitrary position in a moving region.
Another object of the present invention is to provide a position detection system that can detect an arbitrary position in a moving region at a low cost.
Another object of the present invention is to provide a position detection system capable of detecting an arbitrary position while reducing the labor required for arranging the position specifying member in the moving region.

The present invention was devised to achieve the above object, and the self-position detecting device for a moving body according to claim 1 of the present invention comprises at least three retroreflective members and the respective recursive members. A self-position detecting device for a moving body that moves in a moving area having a ceiling on which at least four marks for position identification provided with a base material for fixing a reflecting member are attached in an irregular arrangement, wherein the infrared irradiation unit for irradiating infrared rays toward the ceiling comprising at least four marks, imaging means for imaging at least three marks reflected infrared rays the irradiated, and information regarding the placement of marks in the ceiling , a mark arrangement storing means that associates and stores the position information of the moving region, by image processing the mark which is the imaging, at least corresponding to the retroreflective member The images are clustered and detected as one mark, the center coordinates of the at least three images in the detected marks are calculated as the coordinates of cluster representative points, and the ceiling is obtained from at least three of the captured images. the mark in the number smaller than the total number of marks is adhered to, a mark detection means for detecting as a group characterized by the mutual arrangement of the mark of the number, relating to the arrangement of the mark stored in the mark arrangement storing means With reference to the information , according to the number of cluster representative points in the detected group, all marks attached to the ceiling are divided into at least four group candidates, and the marks of the marks in the group candidates The distance between cluster representative points is obtained as distance data, and each distance data between the cluster representative points in the group candidate is obtained. Data and compares the respective distance data between the clusters representative points in the detected group, said searching the same arrangement as the arrangement of the marks in the detected group, correspondence to the arrangement of the found marked Current position specifying means for specifying the obtained position information as the current position of the moving body.

  According to such a configuration, the self-position detection device of the moving body detects a plurality of marks as a group by imaging a plurality of marks reflecting infrared rays by the imaging unit and performing image processing by the mark detection unit. This group consists of three or more marks. Then, the self-position detecting device specifies the position of the mark constituting the detected group as the current position of the moving body by comparing the detected group with a previously stored group by the current position specifying means. Here, since the marks are irregularly arranged, a group composed of a relatively small number of marks of at least three can be identified. Further, since the current position of the moving body is specified based on a plurality of marks, the position can be detected more accurately than when only one mark is used.

  Further, the mobile body self-position detecting device according to claim 2 is the mobile body self-position detecting device according to claim 1, wherein the mobile body self-position detecting device stores position information of the mobile body according to time. The position information specified by the current position specifying means is stored in the movement locus storage means when the position information management means for storing the position information in the movement locus storage means and the current position cannot be specified by the current position specifying means. And a current position estimating means for estimating the current position of the mobile body based on the position information at the predetermined time and the current time.

  According to such a configuration, the mobile body self-position detection device stores the current position in the movement trajectory storage means by the time by the position information management means, and based on the self-position detected in the past by the current position estimation means. The current position can be estimated. Here, the estimation method is arbitrary, and may be, for example, linear prediction using information of two past points, or curve prediction using information of three or more points in the past.

  The mobile unit self-position detecting device according to claim 3 is the mobile unit self-position detecting device according to claim 2, wherein the location information managing unit is estimated by the current location estimating unit. Based on the position information, the information regarding the mark arrangement stored in the mark arrangement storage means and the position information of the moving area are updated.

  According to such a configuration, the mobile unit's self-position detection device can prioritize the position information estimated by the current position estimation unit over the information regarding the mark arrangement stored in advance in the mark arrangement storage unit. . If the mark is peeled off from the ceiling, the information of the group including the mark cannot be used, so the information stored in advance must be updated. However, the self-position detecting device can automatically update the information related to the mark arrangement stored in the mark arrangement storing means by the position information managing means, so that even if the mark is peeled off, it continues to accurately detect the current position. be able to.

A mobile body self-position detecting device according to claim 4 is the mobile body self-position detecting device according to any one of claims 1 to 3,
Attitude by which the moving body calculates attitude data, which is information on attitudes changed by driving the legs, two legs having a driving mechanism, leg control means for controlling the driving mechanism of the legs A biped walking robot provided with data calculation means, wherein the current position specifying means is based on the posture data and the distance data between the cluster representative points of the marks in the group detected by the mark detection means It is characterized by correcting.

According to such a configuration, the self-position detecting device of the moving body corrects the distance data between the cluster representative points of the marks in the detected group by reflecting the posture of the robot by the current position specifying unit. Therefore, by setting camera parameters such as the position and angle of the imaging means in advance in the posture that serves as the reference for the robot, and acquiring the camera parameters as posture data when the posture of the robot changes, the mark placement can be accurately determined. It can be obtained. As a result, accurate position detection can be performed not only when stationary but also during movement.

A moving body position detection system according to claim 5 is a mobile body self-position detecting device according to any one of claims 1 to 4 and a ceiling of a moving area where the moving body moves. A position detection system comprising at least four position-identifying marks affixed irregularly to each other, wherein the marks fix at least three retroreflective members and each of the retroreflective members. It is characterized by providing the base material for doing.

According to such a configuration, in the position detection system, since the mark includes the retroreflective member, the cost is lower than that of the conventional mark including a member that actively emits light and a storage area that stores predetermined information. Can be reduced. Further, in the position detection system, the self-position detection device clusters at least three images corresponding to at least three retroreflective members and detects them as one mark by the mark detection means. Can be accurately identified.

The position detection system according to claim 6 is the position detection system according to claim 5 , wherein the mark is formed of a transparent or translucent member of the base material. A hole for detaching the mark from the ceiling is provided.

  According to this configuration, in the position detection system, the mark can be peeled off by pulling the hole provided in the mark with a relatively weak force. Further, the peeled mark can be removed cleanly so that it can be reused. Furthermore, since the substrate is transparent or translucent, it does not stand out even after the mark is attached, and the aesthetic appearance is not impaired. The hole provided in the mark is useful, for example, when the mark attachment position is changed and the mark is once removed and then attached again in order to improve the position detection accuracy.

Further, the position detection system according to claim 7 is the position detection system according to claim 5 or 6 , wherein the plurality of marks on the ceiling are simultaneously transmitted to a plurality of positions to be attached. The projector further includes:

  According to such a configuration, in the position detection system, before applying the marks, the projector irradiates light at a plurality of positions to which the plurality of marks are to be attached. The projector has a planar or curved irradiation surface with a plurality of holes. In the position detection system, the more irregular the arrangement of the marks, the easier the group identification of the marks. Therefore, the plurality of holes formed in the irradiation surface are randomly arranged. Further, in this position detection system, since it is necessary to associate the mark arrangement (group) formed as a result of sticking with the mark arrangement (group) to be stored in advance, a plurality of holes formed in the projector Is preferably matched with the projected image of the ceiling. Note that the irradiation surface is preferably a spherical surface or a hemispherical surface.

According to the first aspect of the present invention, since a mark is detected by a group of three or more marks arranged at random, an arbitrary position and orientation can be accurately detected in the moving region.
According to the second aspect of the present invention, by using the position information obtained from the image processing based on the current information and the position information estimated from the past position information, an arbitrary position in the moving region can be obtained. It can be detected accurately.

According to the third aspect of the present invention, even if the mark is peeled off from the ceiling, the current position can be accurately detected continuously.
According to the fourth aspect of the present invention, in the movement area of the robot, accurate position detection can be performed not only when the robot is stationary but also during movement.

According to the fifth aspect of the present invention, it is possible to detect an arbitrary position in the moving region at a low cost.
According to the invention described in 請 Motomeko 6 or claim 7, while reducing the labor required for placement of the mark in the moving area, it can detect any position.

Hereinafter, the best mode (hereinafter referred to as “embodiment”) for carrying out a mobile body self-position detection apparatus and position detection system of the present invention with reference to the drawings is divided into a first embodiment and a second embodiment. Will be described in detail.
First, an overall configuration of a position detection system A including a robot including a position detection unit according to the first embodiment will be described with reference to FIG.

(Configuration of position detection system A)
FIG. 1 is a configuration diagram of a position detection system according to an embodiment of the present invention, where (a) shows an example of a robot and (b) shows an example of a mark.

  As shown in FIG. 1, the position detection system A includes a robot R and a plurality of position identification marks M that are attached in an irregular arrangement on a ceiling F of a moving area in which the robot R moves. Here, the height of the ceiling F is preferably a normal ceiling height, for example, 4 m or less. The interval between the marks M is preferably about the height of the ceiling, for example, about 2 to 5 m. Note that information about the height from the robot R to the mark M (mark height information) is already known.

Here, the robot R is an autonomously moving biped walking robot.
The robot R has a head portion R1, an arm portion R2, and a leg portion R3, and the head portion R1, the arm portion R2, and the leg portion R3 each have a drive mechanism such as an actuator. Driven by the drive mechanism, bipedal walking is controlled by the autonomous movement control unit 50 (see FIG. 2). Details of this bipedal walking are disclosed in, for example, Japanese Patent Application Laid-Open No. 2001-62760. The robot R has at least a body part R4 to which the head part R1, the arm part R2 and the leg part R3 are connected, and a position detection part 70 (see FIG. 2) provided at the back of the body part R4. And a rear housing portion R5.

  In the position detection system A shown in FIG. 1, a projector K is used to attach the mark M to the ceiling F. The projector K irradiates light to a plurality of positions to which a plurality of marks M are to be attached on the ceiling F all at once. The projector K includes an irradiation surface in which a plurality of holes are formed at random in order to project light. This irradiation surface is preferably a hemispherical surface as shown in FIG. Here, the shape of the irradiation surface may be a spherical surface, and may be a flat surface or a curved surface. Moreover, it is preferable that the plurality of holes drilled in the projector K and the projected image of the ceiling are matched. If the projection image is formed on the ceiling F by the projector K, the position where the mark M is to be pasted is clear, so that the labor required for arranging the mark M can be reduced. The projector K is removed from the moving area after the mark M is attached to the ceiling F.

The mark M is configured to be able to reflect infrared rays. For example, as shown in FIG. 1B, the base material B that can be attached to the ceiling and the irradiation surface that is the back of the attachment surface of the base material B And a reflective member disposed on the top. In this embodiment, the shape of the base material B is circular as shown in FIG. In addition, the shape of the base material B is not specifically limited, As shown in FIG.1 (b), a triangle may be sufficient, or a square may be sufficient.
The base material B is preferably composed of a transparent or translucent member, for example, composed of a synthetic resin such as polyethylene or polypropylene. Thus, if it comprises a transparent or semi-transparent member, it will not stand out even after sticking the mark M, and the aesthetic appearance will not be impaired. In addition, the type of the adhesive applied to the sticking surface of the base material B is not particularly limited as long as it becomes transparent when dried, and is, for example, a normal adhesive or glue for sealing.

  The base material B is provided with a hole Q for detaching the mark M from the ceiling. The hole Q is used for pulling when the mark M is peeled off. This is used, for example, when the mark M attachment position is changed in order to improve the position detection accuracy, and when the mark M is once removed and then attached again. As a result, the peeled mark M can be removed cleanly so that it can be reused. The hole Q shown in FIG. 1B is formed integrally with the base material B, but the hole Q may be a separate body from the base material B. For example, a string-like ring made of a transparent or translucent member may be provided at the end of the base material B, and this ring may be used as the hole.

The reflection member H is composed of, for example, a retroreflection member that reflects in the direction of the angle irradiated with infrared rays.
In the present embodiment, three reflecting members H are provided in one mark M. As shown in FIG. 1B, the three reflecting members H are preferably arranged so as to be at the positions of the vertices of the equilateral triangle. When a retroreflective member having a size of about 1 cm square is used as the reflective member H, one side (or diameter) of the base material B is preferably about 5 cm. Further, the number of reflection members H in one mark M may be four or more.

FIG. 2 is a block diagram showing a configuration of the robot shown in FIG.
The robot R is connected to the base station 1 by wireless communication.
The base station 1 is connected to a management computer 3 via a robot dedicated network 2. Although not shown, the management computer 3 is connected to a terminal for inputting commands, data, and the like to the management computer 3 via a predetermined network.
The robot dedicated network 2 connects the base station 1 and the management computer 3 and is realized by a LAN or the like.

  The management computer 3 manages the robot R, and performs various controls such as movement and speech of the robot R via the base station 1 and the robot dedicated network 2 and provides necessary information to the robot R. To do. Here, the necessary information corresponds to a map around the robot R, and the information is stored in a storage unit (not shown) provided in the management computer 3.

Hereinafter, the robot R will be described in detail.
[Robot R]
As shown in FIG. 2, in addition to the head R1, the arm R2, the leg R3, the torso R4, and the rear housing R5, the robot R includes cameras C and C, speakers at appropriate positions of these parts R1 to R5. S, microphone MC, image processing unit 10, audio processing unit 20, storage unit 30, control unit 40, autonomous movement control unit 50, wireless communication unit 60, and position detection unit 70.
Furthermore, the robot R has a gyro sensor SR1 that detects the direction in which the robot R is facing, and a GPS receiver SR2 that acquires the position coordinates where the robot R exists.

[camera]
The cameras C and C are capable of capturing video as digital data. For example, a color CCD (Charge-Coupled Device) camera is used. The cameras C and C are arranged side by side in parallel on the left and right, and the captured image is output to the image processing unit 10. The cameras C and C, the speaker S, and the microphone MC are all disposed inside the head R1.

[Image processing unit]
An image processing unit (image processing means) 10 processes images taken by the cameras C and C, and recognizes surrounding obstacles and people in order to grasp the situation around the robot R from the taken images. Part. The image processing unit 10 includes a stereo processing unit 11a, a moving body extraction unit 11b, and a face recognition unit 11c.
The stereo processing unit 11a performs pattern matching on the basis of one of the two images taken by the left and right cameras C and C, calculates the parallax of each corresponding pixel in the left and right images, and generates a parallax image. The generated parallax image and the original image are output to the moving object extraction unit 11b. This parallax represents the distance from the robot R to the photographed object.

The moving body extraction unit 11b extracts a moving body in the photographed image based on the data output from the stereo processing unit 11a. The reason for extracting the moving object (moving body) is to recognize the person by estimating that the moving object is a person.
In order to extract the moving object, the moving object extraction unit 11b stores images of several past frames (frames), compares the newest frame (image) with the past frames (images), and Pattern matching is performed, the movement amount of each pixel is calculated, and a movement amount image is generated. Then, from the parallax image and the movement amount image, when there is a pixel with a large movement amount within a predetermined distance range from the cameras C and C, it is estimated that there is a person, and as a parallax image of only the predetermined distance range The moving body is extracted, and an image of the moving body is output to the face recognition unit 11c.

The face recognition unit 11c extracts a skin color portion from the extracted moving body, and recognizes the face position from the size, shape, and the like. Similarly, the position of the hand is also recognized from the skin color area, size, shape, and the like.
The recognized face position is output to the control unit 40 as information when the robot R moves and to communicate with the person.

[Audio processor]
The voice processing unit 20 includes a voice synthesis unit 21a and a voice recognition unit 21b.
The voice synthesizer 21a is a part that generates voice data from character information (text data) and outputs the voice to the speaker S based on the speech action command determined and output by the control unit 40. For the generation of the voice data, the correspondence between the character information (text data) stored in the storage unit 30 in advance and the voice data is used. The audio data is acquired from the management computer 3 and stored in the storage unit 30.
The voice recognition unit 21 b receives voice data from the microphone MC, generates character information (text data) from the input voice data, and outputs the character information to the control unit 40. The correspondence relationship between the voice data and the character information (text data) is stored in the storage unit 30 in advance.

[Storage unit 30]
The storage unit 30 is composed of, for example, a general hard disk and stores necessary information (local map data, conversation data, etc.) transmitted from the management computer 3.

[Control unit 40]
The control unit 40 controls the image processing unit 10, the sound processing unit 20, the storage unit 30, the autonomous movement control unit 50, the wireless communication unit 60, and the position detection unit 70.
The control unit 40 includes an attitude data calculation unit 41.
The posture data calculation unit 41 acquires information regarding the arrangement of the head R1, the arm R2, and the leg R3 via the autonomous movement control unit 50, and changes due to driving of the head R1, the arm R2, and the leg R3. Attitude data, which is information about the attitude, is calculated. The calculated attitude data is output to the position detection unit 70.
Although not shown, the control unit 40 has a configuration for executing a predetermined task such as movement or speech based on a task execution command acquired from the management computer 3.

[Autonomous Movement Control Unit]
The autonomous movement control unit 50 includes a head control unit 51a, an arm control unit 51b, and a leg control unit (leg control means) 51c.
The head control unit 51a drives the head R1 according to an instruction from the control unit 40, the arm control unit 51b drives the arm R2 according to the instruction from the control unit 40, and the leg control unit 51c includes the control unit 40. The leg portion R3 is driven according to the instruction.
The data detected by the gyro sensor SR1 and the GPS receiver SR2 is output to the control unit 40 and used to determine the behavior of the robot R.

[Wireless communication part]
The wireless communication unit 60 is a communication device that transmits and receives data to and from the management computer 3. The wireless communication unit 60 includes a public line communication device 61a and a wireless communication device 61b.
The public line communication device 61a is a wireless communication means using a public line such as a mobile phone line or a PHS (Personal Handyphone System) line. On the other hand, the wireless communication device 61b is a wireless communication unit using short-range wireless communication such as a wireless LAN conforming to the IEEE802.11b standard.
The wireless communication unit 60 performs data communication with the management computer 3 by selecting the public line communication device 61 a or the wireless communication device 61 b in accordance with a connection request from the management computer 3.

[Position detector]
The position detection unit (self-position detection device) 70 mainly detects the position where the robot R exists in a room (moving area) where it is difficult to detect the position of the robot R with the GPS receiver SR2. A unit 71 and a top camera 72 are provided.
The infrared irradiation unit (infrared irradiation means) 71 irradiates infrared rays toward a search area, that is, a ceiling F including a plurality of marks M.
The top camera (imaging means) 72 images a plurality of marks M that reflect the irradiated infrared rays.

FIG. 3 is an explanatory diagram showing the arrangement of the infrared irradiation unit and the top camera shown in FIG.
The infrared irradiation unit 71 and the top camera 72 are installed on the upper part of the rear storage unit R5 (see FIG. 1). The height from the top camera 72 to the ceiling F is input in advance to the position detection unit 70 and used for position detection processing.

  The infrared irradiation part 71 is comprised from several LED71a. LED71a irradiates infrared rays. The wavelength of the irradiated infrared ray is, for example, 830 nm. The infrared irradiation unit 71 includes, as the LED 71a, an LED having a high directivity characteristic and an LED having a low directivity characteristic.

The top camera 72 includes a camera body 72a and a camera lens 72b.
The camera body 72a has high infrared sensitivity and has a high resolution of, for example, 1024 × 768 pixels (XGA), and outputs a captured image to the mark detection unit 75 (see FIG. 4).
The camera lens 72b is, for example, a wide angle lens having an angle of view close to 180 degrees. A filter (not shown) is attached to the camera lens 72b. This filter has a characteristic that the wavelength of the infrared ray emitted from the LED 71a has a peak, and the half value of the peak is relatively narrow.

  In the present embodiment, the plurality of LEDs 71a are doubly arranged around the camera lens 72b as shown in FIG. Of the plurality of LEDs 71a arranged in a double manner, a plurality of LEDs having low directivity are arranged around the camera lens 72b, and outside the LED group having low directivity. Around the periphery, LED groups having highly directional characteristics are arranged. Thereby, the infrared light projection characteristics can be made uniform (uniform). Here, the plurality of LEDs 71a are not limited to being double, and may be arranged around the camera lens 72b in a triple or more manner.

FIG. 4 is a block diagram showing an example of the configuration of the position detection unit shown in FIG.
In addition to the infrared irradiation unit 71 and the top surface camera 72, the position detection unit 70 includes an irradiation control unit 73, an imaging control unit 74, a mark detection unit 75, a mark arrangement storage unit 76, and a current position specifying unit 77. And a search range setting unit 78.

The irradiation control unit 73 controls the infrared irradiation unit 71 so that the search range set by the search range setting unit 78 is irradiated with infrared rays.
The imaging control unit 74 controls the top camera 72 so that the search range set by the search range setting unit 78 is imaged. The imaging control unit 74 is synchronized with the irradiation control unit 73, and each time the infrared irradiation unit 71 irradiates infrared light, the top surface camera 72 can image the search range. The camera body 72a (see FIG. 3) of the top camera 72 may be configured to have the function of the imaging control unit 74.

  The mark detection unit (mark detection means) 75 detects a plurality of marks as a group characterized by the arrangement of three or more marks by performing image processing on the captured mark M (mark detection). Process). Details of the mark detection unit 75 will be described later.

  The mark arrangement storage unit (mark arrangement storage means) 76 stores a group of a plurality of marks M as information related to the arrangement of the mark M on the ceiling F in association with the position information of the moving area.

The current position specifying unit (current position specifying unit) 77 refers to the information on the arrangement of the mark M stored in the mark arrangement storage unit 76, and selects a group having the same arrangement as the mark arrangement in the detected group. The position information associated with the searched group is specified as the current position of the robot R. The current position specifying unit 77 outputs the mark ID of the mark M belonging to the search result group to the search range setting unit 78. Further, the current position specifying unit 77 corrects the distance data regarding the arrangement of the marks in the group detected by the mark detection unit 75 based on the posture data output from the posture data calculation unit 41. Details of the current position specifying unit 77 will be described later.

  The search range setting unit 78 sets a search range by calculating a search range, which is a region to irradiate infrared rays, based on the mark ID output from the current position specifying unit 77, and the irradiation control unit 73 and imaging control. This is output to the unit 74. The search range setting unit 78 refers to the mark arrangement storage unit 76 via the current position specifying unit 77 when calculating the search range.

  In the present embodiment, the search range setting unit 78 controls the start and end of the position detection process based on information input (stored) in advance. For example, when a search range as an initial value is input (stored) in advance, the process is started according to a predetermined schedule, and the process is performed when it is determined that an end condition input (stored) in advance is satisfied. finish. Here, the end condition is determined based on, for example, time and position. Even if the initial value (first search range) is not input (stored), the processing may be started by calculating the search range from the mark ID output by the current position specifying unit 77. Good.

FIG. 5 is a block diagram showing the configuration of the mark detection unit and the current position specifying unit shown in FIG.
<Mark detection unit>
As shown in FIG. 5, the mark detection unit 75 includes an image correction processing unit 101, a noise removal unit 102, a labeling unit 103, a mark feature extraction unit 104, a mark area specifying unit 105, a clustering unit 106, And a cluster representative point calculation unit 107.
Based on the output from the top camera 72, the image correction processing unit 101 corrects the input image due to distortion of the camera lens 72b, and then outputs it to the noise removal unit 102.

  The noise removing unit 102 generates a binary image that is an image obtained by binarizing the infrared image output from the image correction processing unit 101. As a result, the background noise can be removed by separating the region (foreground image) that seems to be the mark M from the background image. Here, as a method for generating a binary image, for example, “1” is assigned when a value related to a pixel of the input image is equal to or larger than a predetermined threshold value, and “0” is given otherwise. For example, “1” may be given when the number of pixels existing in an area where pixels having a luminance of a predetermined value or more are connected is, for example, 70 pixels or more. The binary image generated by the noise removal unit 102 is output to the labeling unit 103 and the mark feature extraction unit 104.

  The labeling unit 103 identifies identification information (label) for identifying an area (image object) generated by connecting adjacent pixels in the foreground image of the binary image from which background noise has been removed by the noise removing unit 102. ) Is given (labeled). The label given by the labeling unit 103 is output to the mark feature extraction unit 104. The image object is output to the clustering unit 106.

  The mark feature extraction unit 104 extracts, as an image feature amount, a mark feature for extracting an image object suitable as the mark M from the input binary image. As shown in FIG. And a density extraction unit 112 and a shape extraction unit 113.

The size extraction unit 111 outputs the number of pixels of each image object included in the input binary image to the mark area specifying unit 105 together with a label.
The density extraction unit 112 outputs a ratio (density) of the number of pixels in a predetermined minute region of each image object included in the input binary image to the mark region specifying unit 105 together with a label.
The shape extraction unit 113 outputs the aspect ratio (aspect ratio) of each image object included in the input binary image together with the label to the mark region specifying unit 105.

  The mark area specifying unit 105 marks an image object in which the feature amount (number of pixels, density, and aspect ratio) of each image output from the mark feature extraction unit 104 is within a predetermined threshold range. It is specified as a candidate for a region where M exists (mark region). That is, the mark area specifying unit 105 filters the image object and extracts a plurality (9 or more) of mark areas. Here, for example, an image object having 4 to 50 pixels, a density of 0.4 or more, and an aspect ratio of 0.8 to 1.2 is preferably used as the mark area. . The label of the specified mark area and the number of specified mark areas (number of marks) are output to the clustering unit 106.

  The clustering unit (clustering means) 106 detects a cluster that is a result of clustering three mark regions (images) corresponding to the three reflecting members H as one mark. Here, the cluster means that the size (pixel value) of the mark area is within a predetermined range, the mark areas are within a predetermined distance from each other, and each mark area includes the mark area 3 It is formed by three mark areas in which there are no similar mark areas other than the number of mark areas. Multiple (three or more) clusters are detected. Thereby, even if an incandescent lamp or a fluorescent lamp is reflected in the infrared image as a mark area, it can be removed as noise when compared with the cluster.

In this embodiment, the clustering unit 106, from among all the marks area identified (the number of the n), in order to obtain a mark (clusters) satisfying the condition mark area, _n C ₃ or a combination ( With regard to (Combination), a combination of three mark areas whose distance between the mark areas is a predetermined value (for example, 30 to 50 mm) is defined as a cluster. Here, it is assumed that the number of obtained clusters is k. Note that k is a numerical value of 3 or more.

  The cluster representative point calculation unit 107 calculates the center coordinates of three mark areas respectively included in the k clusters detected as one mark by the clustering unit 106 as cluster representative points. That is, one group of k cluster representative points is obtained. In other words, one group of “k mark candidates” is obtained. This is called a detection group. Information of the detection group formed by the cluster representative point calculation unit 107 is output to the current position specifying unit 77.

<Current position identification part>
As shown in FIG. 5, the current position specifying unit 77 includes a processing target extracting unit 121, a representative point distance calculating unit 122, a group extracting unit 123, and a position information extracting unit 124.

The processing target extraction unit 121 determines the number of permutations of _m P _k based on the information regarding the arrangement of m marks M stored in advance in the mark arrangement storage unit 76 and the number k of cluster representative points. Group candidates are generated and extracted as processing targets. Here, m is the number of marks M installed on the ceiling F of the moving area and registered in advance. The processing target extraction unit 121 assigns a group ID for identification to each of the number of permutation group candidates of _m P _k . Of course, each of these group candidates includes k cluster representative points as in the detection group formed by the cluster representative point calculation unit 107.

The representative point distance calculation unit 122 obtains the mutual distance (representative point distance) of the k cluster representative points included in the group candidate. This representative point distance corresponds to the length of a geometric side determined by determining two cluster representative points. The representative point distance calculation unit 122 calculates the representative point distance for the number of combinations of _k C ₂ for one group candidate. Then, the representative point distance calculation unit 122 calculates a representative point distance for each of the group candidates of all the processing targets (the number of permutations of _m P _k ). Note that the representative point distance calculation unit 122 obtains the representative point distance for each detection group.

  The number of representative point distances described above is an example, and the present invention is not limited to this. Practically, the combination of representative point distances may be enormous. Therefore, the representative point distance calculating unit 122 determines a predetermined mark existing near the target mark from the target mark to the predetermined mark. It is also possible to sort in advance in ascending order of distance and calculate only a predetermined number of combinations. By doing in this way, the load by arithmetic processing can be reduced.

  Further, the representative point distance calculation unit 122 corrects the representative point distance (distance data) based on the posture data input from the posture data calculation unit 41. In the present embodiment, the representative point distance calculation unit 122 determines whether the robot R is stationary based on the posture data of the leg R3 (see FIG. 2) that is input, and the robot R moves. When it is determined that it is in the middle, correction processing that reflects the moving posture with respect to the representative point distance is executed, and when it is determined that it is stationary, the correction processing is not executed. In other words, the distance between the top camera 72 and the ceiling changes in the moving posture, but in such a movement, the change is corrected, and the representative point distance can be accurately obtained. it can. Even when the head is stationary, if the top camera 72 is displaced from the reference position by driving the head R1 (see FIG. 2), a correction process reflecting this may be executed.

  The group extraction unit 123 sets a group candidate whose difference between the representative point distance and the representative point distance of the detected group is equal to or less than a predetermined value (for example, 50 mm) among all group candidates as the only group candidate. Extracted from the mark arrangement storage unit 76. This extracted group candidate is considered to be the same as the detected group. The group IDs of the extracted group candidates, that is, k mark IDs are output to the position information extraction unit 124 and the search range setting unit.

  The position information extraction unit 124 refers to the mark arrangement storage unit 76 and extracts position information (current position information) associated with the acquired group ID (or k mark IDs) as the current position of the robot R. The data is stored in the storage unit 30 (see FIG. 4) via the control unit 40 (see FIG. 4). The stored current position information is used for the action (speech, movement, etc.) of the robot R as necessary. In addition, you may comprise so that current position information may be stored in the memory | storage part which is not shown in the position detection part 70 (refer FIG. 4).

[Robot movement]
In the position detection system A shown in FIG. 1, the robot R is moved by the control unit 40 (see FIG. 2) from the management computer 3 (see FIG. 2) via the wireless communication unit 60 (see FIG. 2). Accompanying task execution instructions and map information are acquired in advance. Then, according to the schedule, the robot R searches for a route from the current position (home position) of the robot R to the task execution position in the movement area (for example, searches for a route between nodes) and moves, executes the task, and ends the task. Each of the route search from the home position to the home position and the movement are sequentially executed. Here, the robot R reaches the destination at the shortest distance while referring to the map data and detecting the current position by the position detection unit 70 (see FIG. 2) when moving.

  Note that the top camera 72 (see FIG. 2) is calibrated in advance for position detection. That is, when the position detection unit 70 (see FIG. 4) captures a calibration image with known coordinate values in an arbitrary coordinate system in the real space (three dimensions), for example, the position of the top camera 72, the camera A conversion matrix for conversion to image coordinates (two-dimensional) is obtained using various camera parameters including the orientation and characteristics of the imaging element and camera lens 72b (see FIG. 3).

(Position detection operation)
Next, the operation of the position detection unit will be described with reference to FIG. 6 (refer to FIG. 4 as appropriate). FIG. 6 is a flowchart showing processing by the position detection unit shown in FIG.
The position detection unit 70 determines whether or not a search range as an initial value is input (stored) by the search range setting unit 78 according to a predetermined schedule (step S1). When the search range as the initial value is not input (stored) (step S1: No), the position detection unit 70 ends the process. On the other hand, when search range information as an initial value is input (step S1: Yes), the search range setting unit 78 outputs the set search range information to the irradiation control unit 73 and the imaging control unit 74. And the position detection part 70 performs control for irradiating infrared rays toward the set search range by the irradiation control part 73, and irradiates infrared rays to a search range (ceiling) by the infrared irradiation part 71 (step) S2). In addition, the position detection unit 70 controls the top camera 72 so that the search range is imaged in synchronization with infrared irradiation by the imaging control unit 74, and the top range camera 72 images the search range (ceiling). (Step S3). Thereby, the position detection part 70 acquires the infrared image of the ceiling containing the mark M which reflected the infrared rays.

  Subsequent to step S2 and step S3, the position detection unit 70 performs mark detection processing by the mark detection unit 75 (step S4), and performs current position specification processing by the current position specification unit 77 (step S5). Details of step S4 and step S5 will be described later. Then, the position detection unit 70 outputs the mark ID to the search range setting unit 78 by the current position specifying unit 77 and stores the specified current position information in the storage unit 30 (see FIG. 2) (step S6). . Subsequently, the position detection unit 70 determines whether or not the end condition is satisfied by the search range setting unit 78 (step S7). If the end condition is satisfied (step S7: Yes), the position detection unit 70 ends the process. On the other hand, when the end condition is not satisfied (step S7: No), the position detection unit 70 newly sets a search range based on the mark ID by the search range setting unit 78 (step S8), and then proceeds to step S2. Return.

  Next, details of the position detection process shown in FIG. 6 will be described with reference to FIG. 7 (see FIG. 5 as appropriate). FIG. 7 is a flowchart showing details of the position detection process shown in FIG. 6, wherein (a) shows the mark detection process and (b) shows the current position specifying process.

(Mark detection processing)
First, with reference to FIG. 7A, the mark detection process in step S4 will be described.
The mark detection unit 75 uses the image correction processing unit 101 to correct the input image based on the distortion of the camera lens 72b based on the output from the top camera 72. And the mark detection part 75 produces | generates a binary image from an infrared image by the noise removal part 102 (step S21). Then, the mark detection unit 75 labels the image object of the binary image by the labeling unit 103 (step S22), and the mark feature extraction unit 104 extracts the mark features related to size, density, and shape as image feature amounts. (Step S23).

  Subsequently, the mark detection unit 75 filters the image object by the mark region specifying unit 105 to specify a plurality of mark regions (step S24), and the clustering unit 106 clusters the specified mark regions (step S25). ). As a result of this clustering, a plurality of (for example, k) clusters formed of three mark regions are detected. Then, the mark detection unit 75 calculates k cluster representative points by the cluster representative point calculation unit 107 (step S26). Thereby, one group (detection group) composed of k cluster representative points is obtained.

(Current position identification processing)
Next, with reference to FIG. 7B, the current position specifying process in step S5 will be described.
Subsequent to step S26 described above, the current position specifying unit 77 causes the processing target extraction unit 121 to extract the processing target of the arithmetic processing based on the number m of marks installed on the ceiling and the number k of cluster representative points ( Step S31). As a result, group candidates as many as the number of permutations of _m P _k are extracted as processing targets. Then, the current position specifying unit 77 calculates the representative point distance determined by selecting two cluster representative points out of k for each group candidate by the representative point distance calculating unit 122 (step S32). . Here, the representative point distance calculation unit 122 corrects the representative point distance by appropriately reflecting the posture data input from the posture data calculation unit 41.

  Subsequently, the current position specifying unit 77 compares each group candidate with the detected group by the group extracting unit 123, and the representative point distance similar to the representative point distance calculated in the detected group from each group candidate. Are extracted (step S33). Thereby, a group having the same shape as the shape formed by the k marks forming the detection group is extracted. Then, the current position specifying unit 77 extracts the position information of the marks belonging to the extracted group from the mark arrangement storage unit 76 by the position information extracting unit 124 (step S34). The current position of the robot R is specified by the position information of the extracted mark, that is, the current position information.

[Specific example of mark detection processing]
FIG. 8 is an explanatory diagram showing an example of the arrangement of marks on the ceiling.
The arrangement of the marks on the ceiling F is indicated by two-dimensional position coordinates. In FIG. 8, the X axis is positive on the left side in the horizontal direction. The Y axis is positive in the vertical direction. A plurality of fluorescent lamps 801 are installed on the ceiling F, and a plurality of marks are irregularly attached. FIG. 8 shows a part thereof. That is, a total of six fluorescent lamps 801 are installed on the ceiling F shown in FIG. 8, three in the X axis direction and two in the Y axis direction, and the fluorescent lamps 801 are not installed. , 14 marks (M1 to M14) are attached.
Hereinafter, an image picked up by paying attention to the four marks (M9 to M12) shown in FIG. 8 will be described with reference to FIG. 9 and FIG. 9 and 10, the mark M is displayed in a direction in which the X axis and the Y axis shown in FIG. 8 are rotated 90 degrees clockwise.

FIG. 9 is an explanatory diagram showing an example of an image used until a mark area is detected.
FIG. 9A is a diagram illustrating an example of an infrared image input to the noise removal unit 102 (see FIG. 5) of the mark detection unit 75. As shown in FIG. 9A, the input image includes four rectangular pixel regions 901 corresponding to the fluorescent lamps. In FIG. 9A, there are twelve minute pixel areas corresponding to the four marks (M9 to M12) shown in FIG. 8 between the two right pixel areas 901. These minute pixel regions correspond to the reflecting member H (see FIG. 1B) provided on the mark. In FIG. 9A, there are six micro pixel areas corresponding to the two marks (M5, M6) shown in FIG. 8 between the two left pixel areas 901. In addition, there are three small pixel regions corresponding to the mark M4 shown in FIG.

  FIG. 9B is a diagram illustrating an example of a binary image binarized by the noise removal unit 102 (see FIG. 5) of the mark detection unit 75. In the binary image, as shown in FIG. 9B, there are four rectangular image objects 902 corresponding to the fluorescent lamps. Further, there are twelve image objects corresponding to the reflecting members H (see FIG. 1) provided on the four marks (M9 to M12) shown in FIG. Similarly, other image objects correspond to the pixel regions shown in FIG.

  FIG. 9C is a diagram illustrating an example of an image labeled by the labeling unit 103 (see FIG. 5) of the mark detection unit 75. The image shown in FIG. 9C is the same as the image object shown in FIG. 9B except that an identification label is assigned to each image object. For example, four rectangular image objects 903 corresponding to fluorescent lamps are given different labels (for example, composed of numbers and character strings) for identifying them. Similarly, a different label for identifying each of the three image objects 904, 905, and 906 corresponding to each reflecting member H of the mark M10 is given. The same applies hereinafter.

  FIG. 9D is a diagram illustrating an example of the mark area specified by the mark area specifying unit 105 (see FIG. 5) of the mark detection unit 75. As a result of filtering by the mark area specifying unit 105 (see FIG. 5), in the image shown in FIG. 9D, there is no longer a rectangular image object corresponding to the fluorescent lamp. The mark area specifying unit 105 (see FIG. 5), for example, marks each of the image objects corresponding to the reflecting members H provided on the four marks (M9 to M12) illustrated in FIG. This is specified as an area 907.

  Even if there is an image object corresponding to the fluorescent lamp as a result of filtering by the mark region specifying unit 105 (see FIG. 5), the fluorescence is obtained by the processing of the clustering unit 106 (see FIG. 5) of the mark detection unit 75. The image object corresponding to the light is removed. The clustering unit 106 (see FIG. 5) detects the dense mark areas 907 shown in FIG. 9D as a cluster. For example, the four marks (M9 to M12) shown in FIG. 8 are detected as four clusters.

FIG. 10 is an explanatory diagram illustrating an example of an image used until the current position is specified.
FIG. 10A shows four clusters (marks M9 to M12) detected by the clustering unit 106 (see FIG. 5) of the mark detection unit 75 (k = 4).
The cluster representative point calculation unit 107 (see FIG. 5) of the mark detection unit 75 obtains the central coordinates of the three reflecting members H provided in the marks M9 to M12 as cluster representative points. That is, a detection group G ₁ composed of k = 4 cluster representative points is obtained.

FIG. 10B is a diagram illustrating an example of a mark arrangement map that matches the arrangement of the marks attached to the ceiling, and information regarding the mark arrangement stored in the mark arrangement storage unit 76 (see FIG. 5). It corresponds to what was visualized. In this example, the mark arrangement map has ten vertices on a boundary line surrounded by a broken line, and a set g ₁ consisting of four vertices connected to each other by a solid line, and 3 to 3 And four vertices respectively connected to five broken lines. These 19 vertices in total correspond to the mark M (when m = 19).

As described above, when k = 4 and m = 14, the processing target extraction unit 121 (see FIG. 5) of the current position specifying unit 77 performs processing for the group candidates corresponding to the number of permutations of ₁₄ P _4. Will be extracted as a processing target. In the mark arrangement map shown in FIG. 10B, only a part of the group candidate is shown as a rectangle.

Then, the representative point distance calculating unit 122 (see FIG. 5) of the current position specifying unit 77 obtains the representative point distance for each group candidate by the number of combinations (6) of ₄ C ₂ . Specifically, the representative point distance calculation unit 122 (see FIG. 5) calculates the lengths of four sides and two diagonal lines of each quadrangle.

Then, (see Fig. 5) group extraction section 123 of the current position determining unit 77 extracts the matched group candidates to detect a group G ₁ shown in Figure 10 (a). For example, in the mark arrangement map shown in FIG. 10B, the sets g ₁ having the most similar lengths of four sides and two diagonal lines are extracted as the same group. Note that pattern matching of the shape itself may be executed.

  According to the first embodiment, the position information in the movement area of the robot R and the group composed of at least three marks M attached to the ceiling F of the movement area are stored in association with each other, and image processing is performed. The current position of the robot R can be specified by identifying the group. In addition, since one mark M is detected as a cluster formed by three image objects, it is possible to detect the position more accurately than in the case where only one image object is used.

(Second Embodiment)
FIG. 11 is a block diagram illustrating another example of the configuration of the position detection unit illustrated in FIG. 4.
As shown in FIG. 11, the position detection unit 70A is provided with a movement locus storage unit 80, a position information management unit 81, and a current position estimation unit 82, except that the position detection unit 70 shown in FIG. Therefore, the same components are denoted by the same reference numerals and description thereof is omitted.

The movement trajectory storage unit (movement trajectory storage means) 80 stores position information of the robot R by time.
The position information management unit (position information management means) 81 is based on the position information estimated by the current position estimation unit 82, information related to the mark arrangement stored in the mark arrangement storage unit 76, position information of the moving area, Is to be updated. Further, the position information management unit 81 stores the position information specified by the current position specifying unit 77 in the movement locus storage unit 80.

The current position estimating unit (current position estimating means) 82 determines whether the current position of the robot R cannot be specified by the current position specifying unit 77 and the position information at a predetermined time stored in the movement locus storage unit 80 and the current time. Based on this, the current position of the robot R is estimated by a known estimation method.
In the present embodiment, the case where the current position of the robot R cannot be specified assumes the case where the mark M has been peeled off from the ceiling F. The known estimation method may be, for example, linear prediction using information on the past two points, or based on curve prediction, which is prediction that associates information on the past three points on a quadratic curve. But you can. In the case of linear prediction, a Kalman filter may be applied. The estimated current position (current position information) is output to the position information management unit 81. The current position specifying unit 77 extracts a mark ID based on the estimated current position information and outputs the mark ID to the search range setting unit 78.

Next, the operation of the position detection unit 70A will be described with reference to FIG. 12 (refer to FIG. 11 as appropriate). FIG. 12 is a flowchart showing processing by the position detection unit shown in FIG.
Each process of step S41 to step S44 executed by the position detection unit 70A is the same as each process of step S1 to step S4 shown in the flowchart of FIG.
Subsequent to step S44, the position detection unit 70A determines whether or not the current position is specified as a result of the current position specifying process performed by the current position specifying unit 77 (step S45). When the current position can be specified (step S45: Yes), the position detection unit 70A stores the specified current position information in the movement trajectory storage unit 80 by the position information management unit 81 (step S46), and the process proceeds to step S49. move on. Here, since each process of step S49 and step S50 is the same as each process of step S7 and step S8 shown in the flowchart of FIG. 6, description is abbreviate | omitted.

  On the other hand, when the current position cannot be specified (step S45: No), the position detection unit 70A is based on the position information of the predetermined time stored in the movement locus storage unit 80 by the current position estimation unit 82 and the current time. The current position of the robot R is estimated (step S47). Subsequently, the position detection unit 70A updates the information stored in the movement locus storage unit 80 with the estimated current position information by the position information management unit 81 (step S48), and proceeds to step S49.

  Since the position detection unit 70A operates as described above, whether or not the mark M has been peeled off the ceiling F by performing an attempt to move the robot R periodically (for example, once a day) on the same course in the movement region. Can be detected. Further, when the answer is No in step S45, the robot R may notify the management computer 3 that the mark M has been peeled off by the wireless communication unit 60 (see FIG. 2).

  Further, as the reason why the current position cannot be specified, there may be a failure of the infrared irradiation unit 71 or the top camera 72 or a malfunction of software for executing image processing or the like. Considering the case where there is a cause on the robot R side as described above, the following operation may be executed. That is, the robot R once returns from the current position (first point) where the position could not be specified to the place (0th point) where the current position (position coordinates) at that time could be specified in the past, and that place (0th point). Point), the position detection unit 70A executes the position detection process again to determine whether or not the current position at that time has been specified. If the position of the 0th point can be specified, the reason why the position could not be detected at the 1st point is likely to be the peeling of the mark M, so the processing from step S47 described above is executed. On the other hand, if the position of the 0th point cannot be specified, it is considered that the reason why the position cannot be detected at the first point is on the robot side. In such a case, for example, the robot R may notify the management computer 3 that a problem has occurred.

  When there are only two or less marks M in the imaging area of the top camera 72, the number of clusters (k) detected by the clustering unit 106 of the mark detection unit 75 as one mark is two. It becomes as follows. For example, when only two marks M can be detected, there is only one representative point distance described above. Therefore, when searching for the same group candidate as the detected group, many similar group candidates are found. As a result, the possibility of misrecognizing a group is considered to increase. Accordingly, such a case may be included as one of cases where the current position of the robot R cannot be specified by the current position specifying unit 77.

  According to the second embodiment, even if the mark M is peeled off from the ceiling F, an unregistered mark M is pasted, or a mark that should not exist is detected, the past Since the current position information can be estimated from the position information, it is possible to continuously detect the current position. Moreover, since the information regarding the mark arrangement stored in the mark arrangement storage unit 76 can be automatically updated, it is possible to save the user from inputting correct information.

  As mentioned above, although preferred embodiment of this invention was described, this invention is not limited to each above-described embodiment. For example, in each embodiment, the position detection unit 70 (70A) has been described as being provided in an autonomous mobile robot capable of walking on two legs. However, the present invention is not limited to this, and the autonomous mobile robot that moves on wheels, industrial Application to various moving bodies such as industrial robots and automobiles is also possible.

It is a block diagram of the position detection system which concerns on embodiment of this invention, (a) is a robot, (b) has shown an example of the mark. It is a block diagram which shows the structure of the robot shown in FIG. It is explanatory drawing which shows arrangement | positioning with the infrared irradiation part shown in FIG. 2, and a top camera. It is a block diagram which shows an example of a structure of the position detection part shown in FIG. It is a block diagram which shows the structure of the mark detection part and current position specific | specification part which were shown in FIG. It is a flowchart which shows the position detection process by the position detection part shown in FIG. FIG. 7 is a flowchart showing details of the position detection process shown in FIG. 6, wherein (a) shows the mark detection process and (b) shows the current position specifying process. It is explanatory drawing which shows the example of arrangement | positioning of the mark in a ceiling. It is explanatory drawing which shows an example of the image used until a mark area | region is detected. It is explanatory drawing which shows an example of the image used until it specifies a present position. FIG. 5 is a block diagram illustrating another example of the configuration of the position detection unit illustrated in FIG. 4. It is a flowchart which shows the process by the position detection part shown in FIG.

Explanation of symbols

A Position detection system R Robot R1 Head R2 Arm R3 Leg R4 Body R5 Back storage F Ceiling M Mark B Base material H Reflective member 1 Base station 2 Robot dedicated network 3 Management computer 10 Image processing unit 20 Audio processing Unit 30 storage unit 40 control unit 41 posture data calculation unit 50 autonomous movement control unit 51a head control unit 51b arm control unit 51c leg control unit (leg control unit)
60 wireless communication unit 70 position detection unit (self-position detection device)
71 Infrared irradiation part (Infrared irradiation means)
71a LED
72 Top camera (imaging means)
72a Camera body 72b Camera lens 73 Irradiation control unit 74 Imaging control unit 75 Mark detection unit (mark detection means)
76 Mark arrangement storage unit (mark arrangement storage means)
77 Current position specifying part (Current position specifying means)
78 Search range setting unit 80 Moving track storage unit (moving track storage means)
81 Location information management unit (location information management means)
82 Current position estimation unit (current position estimation means)
DESCRIPTION OF SYMBOLS 101 Image correction process part 102 Noise removal part 103 Labeling part 104 Mark feature extraction part 105 Mark area | region identification part 106 Clustering part (clustering means)
107 cluster representative point calculation unit 111 size extraction unit 112 density extraction unit 113 shape extraction unit 121 processing target extraction unit 122 representative point distance calculation unit 123 group extraction unit 124 position information extraction unit C camera S speaker MC microphone SR1 gyro sensor SR2 GPS reception vessel

Claims (7)

A moving area having a ceiling on which at least four marks for position identification having at least three retroreflective members and a base material for fixing each retroreflective member are affixed in an irregular arrangement A self-position detecting device for a moving moving body,
Infrared irradiation means for irradiating infrared rays toward the ceiling including the at least four marks;
Imaging means for imaging at least three marks reflecting the irradiated infrared rays;
Mark arrangement storage means for storing information relating to the arrangement of the mark on the ceiling and the positional information of the moving area;
By performing image processing on the captured mark, at least three images corresponding to the retroreflective member are clustered and detected as one mark, and the centers of the at least three images in the detected mark are detected. Coordinates are calculated as the coordinates of cluster representative points, and the number of the captured images and the number of marks less than the total number of marks attached to the ceiling are characterized by the mutual arrangement of the number of marks Mark detection means for detecting as,
With reference to the information on the mark arrangement stored in the mark arrangement storage means, at least four marks attached to the ceiling are selected according to the number of the cluster representative points in the detected group. Dividing into group candidates, each distance of the cluster representative points of each mark in the group candidate is obtained as distance data, each distance data between the cluster representative points in the group candidate, and the detected group Are compared with each distance data between the cluster representative points, and the same arrangement as the arrangement of the marks in the detected group is searched, and the position information associated with the arrangement of the searched marks is obtained. A self-position detecting device for a moving body, comprising: current position specifying means for specifying the current position. Moving trajectory storage means for storing position information of the moving body according to time;
Position information management means for storing the position information specified by the current position specifying means in the movement trajectory storage means;
Current position estimating means for estimating the current position of the moving body based on the position information at a predetermined time stored in the movement locus storage means and the current time when the current position cannot be specified by the current position specifying means. The mobile body self-position detecting device according to claim 1, further comprising: The location information management means includes:
3. The information on the mark arrangement stored in the mark arrangement storage unit and the position information of the moving area are updated based on the position information estimated by the current position estimation unit. The mobile body self-position detecting device according to claim. The moving body has two legs having a driving mechanism, leg control means for controlling the driving mechanism of the leg, and an attitude for calculating attitude data that is information relating to an attitude changed by driving the leg. A biped walking robot equipped with data calculation means,
The current position specifying means corrects distance data between the cluster representative points of the marks in the group detected by the mark detection means based on the posture data. The mobile unit self-position detecting device according to any one of the preceding claims. The mobile body self-position detecting device according to any one of claims 1 to 4, and at least four for position identification attached to a ceiling of a moving area in which the moving body moves in an irregular arrangement . A position detection system comprising a plurality of marks,
The position detection system, wherein the mark includes at least three retroreflective members and a base material for fixing the retroreflective members . The mark, the formed of a base material is transparent or semi-transparent member, the substrate, according to claim 5, characterized in that holes for desorbing the mark from the ceiling are provided The position detection system described in. The position detection system of claim 5 or claim 6, characterized by further comprising a projector that simultaneously irradiated with light in a plurality of positions to be stuck of the plurality of marks each in the ceiling.

Images