JP2009301175A - Monitoring method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a monitoring method of continuously tracking an object which is identified by security monitoring, using a moving robot. <P>SOLUTION: The monitoring method performs: a patrol action that uses a monitor camera 11 and a control camera 12 each having an optical zoom function as cameras to be mounted on the moving robot 10 and that moves the moving robot 10 according to a video image from the control camera 12 while setting the monitor camera 11 in a wide-view patrol state for raster scan; an identification action for zooming in the subjects A, B to be identified while setting the monitor camera 11 in a narrow-view monitoring state if the subjects A, B to be identified are detected in the wide-view video image from the monitor camera 11 during the patrol action; and a tracking action for tracking the objects by moving the moving robot 10 according to the video image from the control camera 12 while keeping the monitor camera 11 in the narrow-view monitoring state if the subjects A, B to be identified are identified as the objects to be tracked. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a monitoring method used when monitoring a specific area using a mobile robot equipped with a camera.

As a mobile robot used in this type of monitoring method, for example, as described in Non-Patent Document 1, there is one equipped with a camera that monitors all directions, a heat source detection camera, and the like. This mobile robot automatically runs on a preset route or is remotely operated based on the image of the camera. Further, as a camera for monitoring omnidirectional, there is a camera that combines a conical omnidirectional mirror and a CCD camera as described in Non-Patent Document 2 as an omnidirectional sensor.
http://plusd.itmedia.co.jp/lifestyle/articles/0407/22/news061.html http://www.vstone.co.jp/top/products/sensor/index.html

  By the way, in the monitoring method using the mobile robot as described above, when the image of the omnidirectional monitoring camera is used for the operation of the mobile robot, it can be easily used as an image (screen) for remote control by the operator, for example. is there.

  However, when using images from an omnidirectional camera for monitoring, a method of projecting and converting the video to digital zoom is used. However, the number of pixels of the camera is limited and a lens is used for data acquisition. Since the compressed information is digitally restored, it may be difficult to identify distant situations.

  For this reason, the conventional monitoring method using a mobile robot has a problem that it is difficult to achieve both a warning action performed by running the mobile robot and a warning monitoring action performed by continuously watching the monitoring video. there were.

  The present invention has been made in view of the above-described conventional situation, and when performing monitoring using a mobile robot, it is possible to achieve both a warning action performed by running a mobile robot and a continuous warning monitoring action. As a result, it is an object to provide a monitoring method capable of continuously tracking a tracking target identified by vigilance monitoring with a mobile robot.

  The monitoring method of the present invention is a method for monitoring a specific area by running a mobile robot equipped with a camera. As a camera mounted on the mobile robot, a monitoring camera having an optical zoom function, and a mobile robot At least using a steering camera that shoots the front, using a surveillance camera with a wide-range surveillance state, and performing a raster scan, while the surveillance action that makes the mobile robot run based on the image from the steering camera, and during the surveillance action, When an identification object is found in a wide-range visual field image from the surveillance camera, the identification camera tracks the identification object by setting the surveillance camera to a narrow-range visual field monitoring state, and the identification object tracks the identification action. Moves based on the video from the control camera while keeping the surveillance camera in a narrow range view when it is identified as the target By traveling bot has a configuration in which the tracking action to track the tracking target, and a means for solving the conventional problems with the above configuration.

  In the monitoring method of the present invention, there are remote control by an operator or automatic control by a computer in order to make the mobile robot run based on an image from the control camera.

  In the surveillance method of the present invention, the surveillance camera has an optical zoom function because detailed data that is continuous in time is necessary to clearly identify an object to be identified in the distance. The zoom function is used to switch between a warning state with a wide visual field and a monitoring state with a narrow visual field. This surveillance camera is held by, for example, a conventionally known pan / tilt gimbal mechanism and performs raster scanning.

  On the other hand, the steering camera is not limited as long as it can capture at least a predetermined range in front of the mobile robot, but in order to properly grasp the situation around the mobile robot, a wide-angle lens such as a fisheye lens or an omnidirectional mirror is used. What you have is more desirable.

  According to the monitoring method of the present invention, when a surveillance object is detected in a wide field of view image from the monitoring camera during a surveillance action in which the mobile robot is run while performing a raster scan with the surveillance camera in a wide field of view surveillance state. As the identification action, the mobile robot is stopped, the monitoring camera is set in a monitoring state with a narrow visual field, and the identification object is closed up.

  When the identification object is identified as the tracking target in the identification action, the mobile robot is moved as the tracking action based on the video from the steering camera while keeping the monitoring camera in a monitoring state with a narrow visual field. Since the vehicle can be run, the tracking target identified by vigilance monitoring can be continuously tracked by the mobile robot.

  According to the monitoring method of the present invention, when performing monitoring using a mobile robot, it is possible to achieve both a warning action performed by running the mobile robot and a continuous warning monitoring action. And, by performing a raster scan with a surveillance camera having an optical zoom function, it is possible to run the mobile robot based on the image of the steering camera while performing warning monitoring with high identification accuracy over a wide area. A very excellent effect is obtained that it is possible to continuously track a tracking target identified by vigilance monitoring with a mobile robot.

  In the monitoring method of the present invention, when the traveling of the mobile robot based on the video from the control camera is a remote control by the operator, and an identification object is found in the wide-range visual image from the monitoring camera A configuration in which an operator operates an image pointing device provided on a monitor dedicated to a monitoring camera to close up an identification target can be adopted. For example, the image pointing device is a touch screen, and the operator uses a touch screen. The structure which closes up an identification target object by touching is employable.

  If the configuration in which the image pointing device is a touch screen is adopted, the man-machine interface becomes extremely simple, and the operator's work load is reduced.

  Furthermore, in the monitoring method of the present invention, it is possible to employ a configuration that automatically closes up an identification object by image processing of a computer when the identification object is found in a wide field of view image from a monitoring camera. .

  When this configuration is adopted, the operator's work load can be greatly reduced.

  In the monitoring method of the present invention, a number of known techniques can be employed as appropriate for the identification and tracking algorithm of the identification object.

  Hereinafter, an embodiment of a monitoring method of the present invention will be described based on the drawings.

  1 and 2 are diagrams for explaining an embodiment of the monitoring method of the present invention. As shown in FIG. 1, this monitoring method includes a monitoring camera (visible monitoring camera and infrared monitoring camera) 11 having an optical zoom function, and a steering camera (manipulation for shooting) at least in front of the mobile robot 10. (Visible camera and steering infrared camera) 12 is mounted on a vehicle-type mobile robot 10, and this mobile robot 10 is run to monitor a specific area. In this embodiment, the operator Op of the control station 1 is operated. The mobile robot 10 is caused to travel by remote control.

  The surveillance camera 11 has an optical zoom function because detailed data that is continuous in time is necessary to clearly identify an identification object in the distance. On the other hand, the steering camera 12 is more preferably equipped with a wide-angle lens such as a fisheye lens or an omnidirectional mirror in order to appropriately grasp the situation around the mobile robot 10.

  The control station 1 that remotely controls the mobile robot 10 includes a control computer 2, a remote control device (including a remote control device for the monitoring camera 11) 3, and a monitor that displays images captured by the monitoring camera 11 of the mobile robot 10. A monitoring monitor 4 and a steering monitor 5 for displaying images captured by the steering camera 12, and the monitoring monitor 4 is provided with a touch screen 6 as an image pointing device.

  The mobile robot 10 includes a surveillance camera 11, a laser distance measuring device, a surveillance unit having a pan / tilt gimbal mechanism for driving them, a steering camera 12, a steering sensor unit having devices such as GPS and various sensors, and a drive A system unit, a power supply unit, and a computer unit are provided. In addition to the in-vehicle computer, the computer unit transmits a transmitter that transmits images captured by the cameras 11 and 12 and vehicle operation data from the operation sensor unit analyzed by the in-vehicle computer and attitude data of the monitoring camera 11. (The laser range finder, pan tilt gimbal mechanism, drive system unit, power supply unit, and computer unit are all not shown).

  In this case, the monitoring camera 11 and the laser distance measuring device of the monitoring unit are arranged at the upper spherical portion of the shaft 21 erected at the center of the vehicle body 20. Further, the steering camera 12 of the steering sensor unit is disposed in the middle portion of the shaft 21, and the steering laser distance measuring device is disposed in the front portion of the vehicle body 20.

  When the mobile robot 10 is run by remote control of the operator Op of the control station 1 and monitoring is performed, first, as shown in FIG. A raster scan is performed in a warning state, and the image is displayed on the monitor 4 for monitoring. The operator Op operates the remote control device 3 while watching the control monitor 5 for displaying the image of the control camera 12 to the mobile robot 10 over the entire area. Is performed (step S1 in FIG. 2).

  During this guarding action, as shown in FIG. 1B, when the identification objects A and B are found in the wide field of view video W from the monitoring camera 11, the mobile robot 10 is operated by the operation of the operator Op. Following the stop, in step S2 of FIG. 2, the operator Op selects either object detection by computer image processing or object selection by the operator Op.

  When the object selection by the operator Op is selected, the identification objects A and B are specified in the wide field-of-view image W in step S3 in FIG. 2, and then the identification is performed in step S4 in FIG. As an action, the operator Op touches the identification object A (or the identification object B) on the touch screen 6 of the monitoring monitor 4, and in step S5 of FIG. Thus, a narrow range visual field image Ta (or narrow range visual field image Tb) in which the identification object A (or identification object B) is close-up is displayed on the monitoring monitor 4 to identify whether or not it is a tracking target.

  Next, in step S6 of FIG. 2, when the identification object A is identified as a tracking target, the monitoring monitor 4 displays a narrow-range visual field image Ta that closes up the identification object A as a tracking action. In other words, while the surveillance camera 11 is kept in a monitoring state with a narrow visual field, the mobile robot 10 is made to track by the remote operation of the operator Op based on the images M1, M2, and M3 displayed on the steering monitor 5. Track the target.

  Thus, in the above-described monitoring method, it is possible to achieve both a warning action performed by running the mobile robot 10 and a continuous warning monitoring action, and after the identification objects A and B are captured, the monitoring monitor 4, the mobile robot 10 can be driven by the remote operation of the operator Op while displaying the narrow-range visual field image Ta in which the identification object A is shown close up. Therefore, the tracking target A identified by vigilance monitoring is continued by the mobile robot 10. Can be tracked. When it is determined that the identification objects A and B are not tracking targets, the surveillance camera is returned to the wide field of view and the security action is resumed.

  In the above-described monitoring method, since the monitoring camera 11 having an optical zoom function is raster-scanned, the mobile robot is based on the image of the control camera 12 while performing warning monitoring with high identification accuracy over a wide area. Can run smoothly.

  Furthermore, in the monitoring method described above, the monitor screen 4 is provided with the touch screen 6 as an image instruction device, and when the identification objects A and B are found in the wide field-of-view video W from the monitoring camera 11, the operator Op. Touches the identification object A (or identification object B) on the touch screen 6 of the monitoring monitor 4 to close up the identification object A (or identification object B). (Steps 2 to 4 and 6 in FIG. 2) are extremely simple, and the workload of the operator Op is reduced.

  In the embodiment described above, when the identification objects A and B are found in the wide field-of-view video W from the monitoring camera 11, in step S2 in FIG. 2, the object is detected by computer image processing or the object is selected by the operator Op. However, if the operator Op selects the object detection by computer image processing, the identification objects A and B are automatically identified by computer image processing in step S7 of FIG. In this case, too, the workload of the operator Op is greatly reduced.

  The monitoring method of the present invention can be applied to patrols of various facilities such as airports, harbors, bases, parks, and various buildings. It can also be applied to situation monitoring actions (reconnaissance actions) that are performed prior to rescue operations of personnel in the event of a fire or occupied by a terrorist group.

  In the monitoring method of the present invention, the mobile robot is made to travel based on the image of the control camera while raster scanning the monitoring camera having an optical zoom function. In addition, it is very effective for surveillance and surveillance over a wide area including not only the ground but also the sky.

It is figure (a), (b) explaining the change of the image | video of the monitor for monitoring from the guarding action of the monitoring method by one Embodiment of this invention to the tracking action of tracking object. It is a block diagram explaining the flow from the guarding action of the monitoring method in FIG. 1 to the tracking action of the tracking target.

Explanation of symbols

1 control station 4 monitor for monitoring 5 monitor for operation 6 touch screen (image instruction device)
10 Mobile Robot 11 Surveillance Camera 12 Steering Camera Op Operator W Wide Field of View Image

Claims (4)

A method of monitoring a specific area by running a mobile robot equipped with a camera,
As a camera mounted on a mobile robot, a surveillance camera having an optical zoom function and a control camera for photographing at least the front of the mobile robot are used.
While performing a raster scan with the surveillance camera in a wide range of surveillance, the surveillance action that causes the mobile robot to travel based on the video from the steering camera,
During the police action, when the identification object is found in the wide field of view image from the monitoring camera, the identification action to close the identification object by setting the monitoring camera to the narrow range visual field monitoring state,
When the identification object is identified as the tracking target in the identification action, the tracking camera is tracked by running the mobile robot based on the video from the steering camera while keeping the monitoring camera in a monitoring state with a narrow visual field. A monitoring method characterized by performing a tracking action.   The traveling of the mobile robot based on the image from the control camera is the operation by the remote control of the operator, and when the identification object is found in the wide field of view image from the monitoring camera, the operator The monitoring method according to claim 1, wherein the identification target is closed up by operating an image instruction device provided on the monitor.   The monitoring method according to claim 2, wherein the image pointing device is a touch screen, and an operator touches the touch screen to close up the identification object.   The monitoring method according to claim 1, wherein when an identification object is found in a wide field-of-view image from a monitoring camera, the identification object is automatically closed up by image processing of a computer.

Images