JP2007249735A - Robot location controller and robot self-location restoration method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a technique for accurately recognizing and restoring a self-location and a direction even when a robot is in an error state and the self-location and the direction are lost. <P>SOLUTION: In the robot 1, when an RFID tag detection part 101 detects an RFID tag by an RFID reader 20, a location specifying part 102 specifies the location range of the robot 1 from the location of the detected RFID tag. When a landmark detection part 104 detects a landmark from an image imaged by a stereo camera 30, a location/direction specifying part 106 specifies the self-location of the robot 1 from the location and direction or the like of the detected landmark. When it is confirmed that the specified self-location of the robot is within the specified location range by a location collation part 108, a location restoration part 111 restores the robot 1 to an initial location or a work continuing location. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a self-localization method for a mobile robot that can move autonomously according to route or map information, and in particular, from an error state, the initial position or work based on global sensor information and visual measurement results of the robot. The present invention relates to a robot position control device and a robot self-position recovery method that enable recovery to a continuous position.

  When a robot that can move autonomously enters a serious error state, it may lose its own position and orientation information on the map or route and cannot move to the target position or initial position. Therefore, it is necessary to be able to correctly recognize and recover from the self-position and orientation even when an error occurs.

  As a prior art in which a mobile robot recognizes a position and a direction, for example, there are those described in Patent Document 1 and Patent Document 2 below.

  In the environment identification device described in Patent Document 1, a plurality of landmarks are arranged, and a robot device that moves in a uniquely identifiable environment identifies a current environment from a plurality of registered environments. An environment for recognizing landmarks in the environment, calculating the movement state quantity of the robot device itself, and constructing a current environment map having the landmark position information in the current environment based on the landmark and movement state quantity Environment that identifies the current environment from the similarity between the registered environment map and the current environment map, the map construction unit, the environment map storage unit having a registered environment map database composed of landmark position information and environment ID It consists of an identification unit and an environment search unit that searches for a new environment.

Further, in the position and direction recognition method of the mobile robot described in Patent Document 2, the absolute coordinates with respect to the current position of the mobile robot are used in order to enable stable position and direction recognition with a fast sampling speed and a limited error range. And get the relative coordinates for the moving displacement, and reflect the relative coordinates to the absolute coordinates to recognize the current position and direction. RFID is used as a method for detecting absolute coordinates, and dead reckoning equipment using a speed sensor for detecting the moving speed of the robot and a direction sensor for detecting the moving direction of the robot is used as a method for detecting the relative coordinates. is doing.
JP 2004-110802 A Japanese Patent Laid-Open No. 2004-21978

  In the technique described in Patent Document 1, an environment map is constructed from the recognition results of a plurality of landmarks, and the current environment is identified from the similarity with the registered environment map. In such a method using an observation model, the accuracy of position identification depends on the parameters of the model, and the position and orientation of the robot may not be recognized accurately depending on the arrangement of landmarks. Also, there is no detailed description on how to recover when the robot enters an error state.

  In the technique described in Patent Document 2, the odometry method (dead reckoning navigation) is used, and measurement errors accumulate when the travel distance of the robot increases. Also, no consideration is given to how to recover when the robot enters an error state.

  SUMMARY OF THE INVENTION It is an object of the present invention to solve the above problems and to recognize and recover them accurately even when the robot is in an error state and loses its own position and orientation.

  In order to solve the above problems, the present invention performs global position detection using an RFID system when a robot enters an error state, and identification of a position and orientation by landmark stereo measurement using a stereo camera. Do.

Specifically, the present invention uses the following means in a mobile robot having a stereo camera and a head that rotates horizontally and vertically.
(1) An RFID system is arranged in the external environment of the robot. The robot has an RFID reader.
(2) Place a planar landmark on the floor or wall in the external environment. The landmark uses a symbol or pattern that can display a number and orientation registered in advance, and recognizes the landmark in the image captured by the camera so that both the absolute position and the direction can be grasped.
(3) When the robot loses its own position, the self position and direction are identified as follows.

 (3-1) Obtain the current position using the RFID system.

 (3-2) Rotate the head and search for landmarks with a stereo camera.

 (3-3) When the search is successful, re-check the registered landmark position and the RFID result read by the RFID reader. If the result is valid, it uses the landmark direction and relative position information to move to the landmark position and identify its own position.

(3-4) Resume work from the identified position. Or return to the initial position.
(4) If the search fails, the administrator is called by lighting the lamp or issuing a warning sound. The manager uses the landmark numbers nearby to instruct the robot on position and orientation information. As a result, the robot automatically moves to the landmark position and identifies its own position.

  Specifically, the present invention provides a robot equipped with a moving means for autonomously moving, an RFID reader that reads information from an RFID tag, and a stereo camera that captures a stereo image when the robot loses its own position. An RFID tag information storage means for storing correspondence information between the information of a plurality of RFID tags arranged in the movement range of the robot and the positions of the RFID tags; The position of the RFID tag detected based on the information of the RFID tag detected by the RFID reader and the landmark information storage means for storing the identification information for uniquely identifying the landmark and the corresponding information of the position of the landmark Is obtained from the RFID tag information storage means, and the range of the robot's own position is determined based on the acquired RFID tag position. Position range specifying means for specifying a position range, landmark detection means for detecting a landmark from a stereo image captured by a stereo camera, and a landmark detected as a robot from a stereo image in which the detected landmark is reflected Based on the identification information of the detected landmark and the distance measuring means for measuring the distance between the two, the position of the detected landmark is acquired from the landmark information storage means and described in the detected landmark The direction of the robot is determined based on the information indicating the calculated direction, and the robot's self-position is determined based on the position of the acquired landmark, the determined direction, and the distance between the measured landmark. The self-position specifying means to be identified is compared with the specified position range and the specified self-position, and it is determined that the specified self-position is within the specified position range. If it is, it characterized in that it comprises a position matching means for identifying the identified self-position.

  This makes it possible to perform highly accurate and reliable self-position recovery without administrator intervention.

  Further, the present invention provides an alarm output means for outputting an alarm indicating that the self-position cannot be identified when the self-position cannot be identified by the position collating means in the above robot position control device, and the position is inputted by the administrator. And a position input means for identifying the input position as a self-position.

  Further, according to the present invention, in the above robot position control device, the robot is moved to a predetermined initial position or a work position to be performed next by controlling the moving means based on the identified self position. It is characterized by comprising position recovery means for moving.

  According to the present invention, when the robot is in an error state, the self-position is identified based on the external global sensor information and the visual measurement result of the robot, and is restored to the initial position (or work continuation position). Therefore, the robot can return to the initial position without human intervention as an administrator. In addition, since multiple sensors are used in combination and the self-location is identified by stereo measurement using the vision of the robot, accuracy and reliability are high. By recovering the self-position, continuous autonomous running control becomes possible. Or you can continue a given task.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a diagram showing a functional configuration example of a robot according to an embodiment of the present invention. The robot 1 includes a robot position control unit 10, an RFID reader 20, a stereo camera 30, an operation panel 40, a motor 50, and a movement control unit 51.

  FIG. 2 is a diagram illustrating a hardware configuration example of the robot according to the present embodiment. The robot 1 includes a CPU 11, a memory 12, an RFID reader 20, a stereo camera 30, an operation panel 40, a motor 50, and various sensors 60.

  The robot position control unit 10 is a computer including a CPU 11 and a memory 12 and controls the position of the robot 1. The robot position control unit 10 includes an RFID tag detection unit 101, a position specification unit 102, an RFID tag information storage unit 103, a landmark detection unit 104, a stereo measurement processing unit 105, a position / An azimuth specifying unit 106, a landmark information storage unit 107, a position collation unit 108, an alarm output unit 109, a position input unit 110, and a position recovery unit 111 are provided.

  The RFID tag detection unit 101 detects the RFID tag by acquiring the tag ID read from the RFID tag by the RFID reader 20.

  The position specifying unit 102 refers to the RFID tag information storage unit 103 with the tag ID of the RFID tag detected by the RFID tag detection unit 101, and acquires the position of the RFID tag. Based on the acquired position of the RFID tag, the position range which is the range of the robot 1's own position is specified.

  FIG. 3 is a diagram illustrating an example of a data structure of the RFID tag information storage unit. The RFID tag information storage unit 103 stores correspondence information between the tag ID of the RFID tag and the position where the RFID tag is installed.

  The landmark detection unit 104 detects a landmark from the stereo image captured by the stereo camera 30.

  The stereo measurement processing unit 105 measures the distance from the robot 1 to the landmark based on the principle of triangulation based on the stereo image in which the landmark is reflected.

  The position / orientation specifying unit 106 refers to the landmark information storage unit 107 with the landmark identification mark detected by the landmark detection unit 104 and acquires the position of the landmark. Further, the direction is obtained from the direction mark of the landmark detected by the landmark detection unit 104. Based on the position and orientation of the acquired landmark and the distance from the robot 1 to the landmark measured by the stereo measurement processing unit 105, the self position of the robot 1 is specified.

  FIG. 4 is a diagram illustrating an example of a landmark. The landmark according to the present embodiment is composed of an identification mark (numerical number in the figure) for uniquely identifying the landmark, and an orientation mark (direction indication of a triangle in the figure) indicating the orientation. Here, it is assumed that the triangle indicates the north direction when facing upward, and the robot 1 that captures the landmark obtains the direction in which the triangle indicates the direction in which the robot is facing. To do. For example, when the robot 1 captures the landmark of FIG. 4 from the front, the triangle mark indicating the azimuth is pointing to the right, so that the azimuth that the robot 1 is currently facing can be determined to be east.

  FIG. 5 is a diagram illustrating an example of a data structure of the landmark information storage unit. The landmark information storage unit 107 stores correspondence information between an identification mark for uniquely identifying a landmark and a position where the landmark is installed. In FIG. 5, a pattern indicating a landmark number is shown as an identification mark, but in practice, the landmark information storage unit 107 stores an identification number of a pattern recognition result.

  The position collating unit 108 collates the position range of the robot 1 specified by the position specifying unit 102 with the self position of the robot 1 specified by the position / orientation specifying unit 106, and is specified by the position / orientation specifying unit 106. It is determined whether the self position of the robot 1 is within the position range of the robot 1 specified by the position specifying unit 102. If the robot 1 specified by the position / orientation specifying unit 106 is within the position range of the robot 1 specified by the position specifying unit 102, the robot 1 specified by the position / orientation specifying unit 106 Identify self-location.

  When the self-position of the robot 1 cannot be identified, the alarm output unit 109 notifies the manager to that effect. As a method of notifying the administrator, for example, various methods such as notifying the administrator's server that the self-position is not detected, turning on a warning lamp, and sounding an alarm sound are conceivable.

  The position input unit 110 inputs information on the position and direction designated by the administrator. As a method of instructing the position and direction of the robot by the administrator, for example, a method of instructing by remote operation from the administrator's server using a wireless LAN, a method of instructing directly by the operation panel, or the like can be considered.

Here, as a procedure when using a wireless LAN, for example,
1) When the robot 1 cannot identify its own position, it acquires a surrounding image,
2) Notifying the administrator's server that the self-location cannot be identified by transmitting the acquired surrounding image and the scenario ID for specifying the currently executed scenario using the wireless LAN,
3) The administrator identifies the position of the robot 1 by comparing the feature points transferred to the image around the robot 1 sent to the server with the map data corresponding to the scenario ID sent to the server. Then, a procedure is used in which the server 1 is instructed to the robot 1 of the position and orientation information using the wireless LAN.

As a procedure when using the operation panel 40, for example,
1) When the robot 1 cannot identify its own position, it calls an administrator by sounding an alarm,
2) The administrator displays a map of the scenario currently being executed on the operation panel 40 of the robot 1,
3) The administrator instructs the robot 1 on position and orientation information from the operation panel 40 with reference to the landmarks registered in advance.
The following procedure is used.

  The position recovery unit 111 issues a movement instruction to the movement control unit 51 that controls the motor 50 based on the obtained information on the self position and direction of the robot 1 and returns the robot 1 to the initial position. The robot 1 may be moved to the work continuation position without returning to the initial position, and the work may be resumed.

  FIG. 6 is a diagram for explaining an outline of self-position recovery of the robot according to the present embodiment. If the robot 1 enters a serious error state and loses information such as its position and direction, it cannot move to the target position or initial position of the work. At this time, the robot 1 automatically attempts to recover its own position.

  First, the robot 1 detects the tag ID from the surrounding RFID tag 70 by the RFID reader 20. With reference to the RFID tag information storage unit 103, the position of the RFID tag 70 having the detected tag ID is acquired. Since the reading range of the RFID reader 20 is limited, in order to detect the tag ID from the RFID tag 70, the robot 1 must be within a certain distance from the RFID tag 70. That is, if the tag ID can be detected, the robot 1 exists within a certain distance from the RFID tag 70 having the tag ID. Therefore, the approximate position range of the robot 1 (the range of the broken semicircle in the figure) can be specified from the position of the RFID tag 70 having the detected tag ID.

  Further, the robot 1 images the surroundings with the stereo camera 30 and detects the landmark 71 from the captured image. With reference to the landmark information storage unit 107, the position of the detected landmark 71 is acquired. Since the captured image is a stereo image, the distance from the robot 1 to the landmark 71 can be measured. Further, from the mark indicating the orientation of the landmark 71, it is possible to determine in which orientation the landmark 71 is located when viewed from the robot 1. Therefore, the self position of the robot 1 can be identified from the information on the position, distance, and direction of the landmark 71 in the captured image.

  The robot 1 collates the position range specified from the position of the RFID tag 70 with the self-position specified from the position of the landmark 71, and the self-position specified from the position of the landmark 71 is the position of the RFID tag 70. If it is within the position range specified from the above, it is determined that the self-position with high accuracy is specified.

  When the RFID tag 70 is not detected, the landmark 71 is not detected, or when the accuracy of the specified self-position is low, the robot 1 is moved to detect the RFID tag 70 and the landmark 71 Try again to detect and locate itself. When the self-position of the robot 1 cannot be specified until the specified condition is satisfied, the fact is notified to the administrator. The prescribed conditions are, for example, conditions such as “5 minutes from the start of self-position recovery”, “up to 5 retries”, and the like. Upon receiving the notification from the robot 1, the administrator inputs position and direction information to the robot 1 by the method using the wireless LAN or the operation panel 40 described above.

  The robot 1 moves to the initial position when its own position and direction are specified. Alternatively, the work may be resumed by moving to the work continuation position without moving to the initial position.

  FIG. 7 is a flowchart of robot self-position recovery processing. The flowchart in FIG. 7 is a flowchart when the robot 1 recovers its own position and then moves to the initial position.

  When the robot 1 loses its own position information due to an error or the like, first, the RFID tag 20 detects the RFID tag (step S10).

  If the RFID tag is detected (step S11), the position of the detected RFID tag is acquired from the RFID tag information storage unit 103 based on the tag ID of the detected RFID tag (step S12). A position range is specified from the acquired position of the RFID tag (step S13).

  Next, the robot 1 performs a landmark detection process (see FIG. 8 described later) (step S14). If the landmark is detected (step S15), the distance from the image in which the detected landmark is shown to the landmark is measured (step S16). The self position is specified from the distance to the detected landmark and the position and orientation of the detected landmark (step S17).

  The position range specified from the RFID tag and the self-position specified from the landmark are collated (step S18). If the specified self-position is within the specified position range (step S19), the specified position range is specified. The self-position is identified (step S20), the robot 1 moves from the identified self-position to the initial position (step S21), and the process ends.

  If the RFID tag cannot be detected in step S11, a landmark detection process is performed (step S22). If the landmark is detected (step S23), the distance from the image in which the detected landmark is reflected to the landmark is measured (step S24). The self position is specified from the distance to the detected landmark and the position and orientation of the detected landmark (step S25).

  It is determined whether or not the self-position search has failed (step S26). If the self-position search has failed, a notification is made to the administrator that the self-position could not be found by sounding an alarm or the like (step S26). S27), receiving an instruction of the position and orientation from the manager (step S28), identifying the designated position as a self-position (step S20), moving from the identified self-position to the initial position (step S21), The process ends. If the search for the self position is not a failure, the robot 1 corrects the position (step S29), returns to step S10, and tries to recover the self position again. At this time, if the self-position can be specified in step S25, the position is corrected based on that.

  If the landmark cannot be detected in step S15, it is determined whether or not the self-position search has failed (step S26). If the self-position search has failed, an alarm is sounded to the administrator. (Step S27), receives an instruction of the position and direction from the administrator (step S28), identifies the designated position as a self-position (step S20), and identifies the identified self The position is moved from the position to the initial position (step S21), and the process is terminated. If the search for the self position is not a failure, the robot 1 corrects the position (step S29), returns to step S10, and tries to recover the self position again.

  In step S19, if the specified self-position is not within the specified position range, it is determined whether the self-position search has failed (step S26). If the self-position search fails, an alarm is sounded, etc. To notify the administrator that the self-position could not be searched (step S27), receives an instruction of the position and orientation from the administrator (step S28), and identifies the instructed position as the self-position (step S28). In step S20), the identified position is moved to the initial position (step S21), and the process is terminated. If the search for the self position is not a failure, the robot 1 corrects the position (step S29), returns to step S10, and tries to recover the self position again. At this time, the position is corrected based on the self-position specified in step S17.

  The determination of the self-location search failure is performed by determining whether a prescribed condition such as “up to 5 retries” is satisfied.

  FIG. 8 is a landmark detection processing flowchart. The robot 1 detects the landmark from the image captured by the stereo camera 30 (step S30).

  If the landmark is detected (step S31), the detected landmark position information is acquired from the landmark information storage unit 107 based on the detected landmark identification mark (step S32). The orientation is obtained from the orientation mark of the landmark (step S33), and the process is terminated.

  If the landmark cannot be detected (step S31), the stereo camera 30 is rotated by pan / tilt (step S34). If the landmark detection is not failed (step S35), the process returns to step S30, and the landmark is detected again. Try detection. If the landmark detection fails (step S35), it is determined that the landmark has not been detected, and the process ends.

  Note that the landmark detection failure is determined by determining whether a prescribed condition such as “up to 5 retries” is satisfied.

  The robot self-position recovery processing as described above can be realized by a computer built in the robot and a software program recorded on a recording medium readable by the computer.

It is a figure which shows the function structural example of the robot by embodiment of this invention. It is a figure which shows the hardware structural example of the robot by this Embodiment. It is a figure which shows the example of the data structure of a RFID tag information storage part. It is a figure which shows the example of a landmark. It is a figure which shows the example of the data structure of a landmark information storage part. It is a figure explaining the outline | summary of the self-position recovery of the robot by this Embodiment. It is a robot self-position recovery process flowchart. It is a landmark detection process flowchart.

Explanation of symbols

1 Robot 10 Robot Position Control Unit 11 CPU
DESCRIPTION OF SYMBOLS 12 Memory 20 RFID reader 30 Stereo camera 40 Operation panel 50 Motor 51 Movement control part 101 RFID tag detection part 102 Position specification part 103 RFID tag information storage part 104 Landmark detection part 105 Stereo measurement processing part 106 Position / azimuth specification part 107 Land Mark information storage unit 108 Position verification unit 109 Alarm output unit 110 Position input unit 111 Position recovery unit

Claims (5)

Has a function of recovering its own position when a robot equipped with a moving means for autonomously moving, an RFID reader that reads information from an RFID tag, and a stereo camera that captures a stereo image loses its own position A robot position control device,
RFID tag information storage means for storing correspondence information between information of a plurality of RFID tags arranged in the movement range of the robot and the position of each RFID tag;
Landmark information storage means for storing correspondence information between identification information for uniquely identifying the landmark and the position of the landmark;
Based on the information of the RFID tag detected by the RFID reader, the position of the detected RFID tag is acquired from the RFID tag information storage means, and the robot's self is acquired based on the acquired position of the RFID tag. A position range specifying means for specifying a position range which is a position range;
Landmark detection means for detecting a landmark from a stereo image captured by the stereo camera;
Distance measuring means for measuring a distance between the robot and the detected landmark from a stereo image including the detected landmark;
Based on the identification information of the detected landmark, the position of the detected landmark is acquired from the landmark information storage means, and based on the information indicating the orientation described in the detected landmark. Self-position specifying means for determining the self-position of the robot based on the obtained position of the landmark, the obtained direction and the distance between the measured landmark and ,
The specified position range is compared with the specified self-position, and when it is determined that the specified self-position is within the specified position range, the specified self-position is identified. A robot position control device comprising: a position matching unit. The robot position control device according to claim 1,
An alarm output means for outputting an alarm indicating that the self-position cannot be identified when the self-position cannot be identified by the position matching means;
A robot position control device comprising: position input means for identifying the input position as a self-position when the position is input by an administrator. In the robot position control device according to claim 1 or 2,
And a position recovery means for moving the robot to a predetermined initial position or a work position to be performed next by the robot by controlling the moving means based on the identified self-position. Robot position control device. Movement means for autonomously moving, an RFID reader that reads information on the RFID tag, a stereo camera that captures a stereo image, information on a plurality of RFID tags arranged in the movement range of the robot, and information on each RFID tag A robot having RFID tag information storage means for storing correspondence information with a position, and landmark information storage means for storing identification information for uniquely identifying a landmark and correspondence information with the position of the landmark is self-positioning. Robot self-position recovery method performed when loss of
The robot
Detecting RFID tag information by the RFID reader;
Acquiring the position of the detected RFID tag from the RFID tag information storage means based on the information of the detected RFID tag;
Identifying a position range that is a range of the robot's own position based on the position of the acquired RFID tag;
Detecting a landmark from a stereo image captured by the stereo camera;
Acquiring the position of the detected landmark from the landmark information storage means based on the identification information of the detected landmark;
A process for obtaining an orientation based on information indicating the orientation described in the detected landmark;
Measuring a distance between the robot and the detected landmark from a stereo image including the detected landmark;
Identifying the position of the robot on the basis of the position of the acquired landmark, the determined azimuth, and the distance between the measured landmark;
The specified position range is compared with the specified self-position, and when it is determined that the specified self-position is within the specified position range, the specified self-position is identified. A robot self-position recovery method characterized by comprising a process. The robot self-position recovery method according to claim 4,
A position recovery process for moving the robot to a predetermined initial position or a work position to be performed next by the robot by controlling the moving means based on the identified self-position; Robot self-position recovery method.

Images