KR20110054614A - Mobile system for recogning position of mobile device and method for estimating position - Google Patents

Abstract

PURPOSE: A mobile system for recognizing position of a mobile device and a method for estimating position are provided to improve the position of a moving object and the accuracy of designing a map by inputting the images of the ceiling and wall through one camera. CONSTITUTION: In a mobile system for recognizing position of a mobile device and a method for estimating position, the images of the ceiling and wall are obtained at the same time(S410). A feature point and a feature line are extracted from the obtained image(S420). A method of estimating a moving object is selected(S430).

Description

MOBILE SYSTEM FOR RECOGNING POSITION OF MOBILE DEVICE AND METHOD FOR ESTIMATING POSITION}

The present invention relates to a movement system and a position estimating method for recognizing the position of a moving object traveling indoors. It is about an estimation method.

Moving objects that perform various functions such as cleaning and security monitoring require accurate location recognition for safe and accurate automatic driving in indoor spaces. Moving object position recognition methods have been attempted using various sensors, and position recognition by camera imaging apparatuses having excellent performance compared to sensor prices is increasing.

Recently, a location recognition method using a ceiling feature point for generating a driving map by recognizing the ceiling feature point with a camera and estimating the position of the robot has attracted attention due to the convenience of implementation. In the ceiling position recognition method, since the ceiling is generally present on a floor and a predetermined height in an indoor space, a driving map can be easily generated using only two-dimensional image processing of a ceiling feature point input by one camera. However, in the conventional ceiling position recognition method, even if the ceiling camera uses a wide-angle lens, the recognition range of the robot is limited because the camera recognition range is limited, and the position error of the moving object occurs when the ceiling feature point does not exist within the recognition range. Therefore, it is difficult to apply in the indoor space area where the ceiling feature points are monotonous or wider than the camera recognition area.

The present invention provides a mobile system and method capable of simultaneously receiving a ceiling and a wall omnidirectional image through a single camera to improve the position of the moving object and the accuracy of the map making and to extend the recognition range.

In addition, the present invention provides a mobile system and method for receiving a ceiling and a wall omnidirectional image simultaneously through a single camera to generate a reliable position and driving map of a moving object in a large indoor space.

The present invention also provides a mobile system and a method for generating a driving map by estimating the position of a moving object by tracking a vertical line of a wall in an area having little or no ceiling feature points by simultaneously receiving a ceiling and a wall omnidirectional image through a single camera. .

According to an exemplary embodiment of the present invention, a moving system includes a driving unit for driving a moving body, a sensor unit providing control information for controlling the driving unit, and a ceiling installed in a vertical direction from a floor to capture a ceiling / wall omnidirectional image. Estimating the position of the moving object by using feature information of the ceiling / wall omnidirectional image captured by the wall omnidirectional image input device and the ceiling / wall omnidirectional image input unit, and controlling the driving unit using the estimated position information. It includes a moving body control unit.

According to the present invention, the ceiling / wall omnidirectional image input device includes a convex reflection mirror having a hole in the center, a wide-angle lens mounted at the center hole of the reflection mirror, and a lens support for positioning the reflection mirror at a predetermined distance from the camera. The lens adapter and the lens adapter is mounted to the front, including a camera for capturing the ceiling / wall omnidirectional image, and is installed on the upper end of the movable body in a direction perpendicular to the bottom surface.

In addition, the position estimation method according to an embodiment of the present invention is an image acquisition step of simultaneously acquiring the ceiling and the wall omnidirectional image through the ceiling / wall omnidirectional image input means, and the feature points and feature lines from the obtained ceiling and wall omnidirectional image An image feature extraction step of extracting and a position estimation step according to an environment for estimating the position of the moving object by selecting a position estimation method of the moving object according to the extracted feature.

In addition, in one aspect of the present invention, the image feature extraction step includes detecting a vertical feature line converging to the center of the acquired ceiling image, and tracking the vertical feature line according to a change in the position of the moving object to obtain a strong vertical feature line. Verifying, correcting distortion of the wide-angle lens of the ceiling area, extracting ceiling feature points from the corrected ceiling image, and generating moving object position feature data comprising the extracted wall vertical feature lines and the ceiling feature points. It may include a step.

In addition, in one aspect of the present invention, the position estimation step according to the environment may estimate the position of the moving object by tracking the ceiling feature point when the ceiling feature point exists, and the azimuth angle of the wall vertical feature line when the ceiling feature point does not exist. The change amount can be tracked to estimate the position of the moving object.

In addition, verifying the vertical feature line in one aspect of the present invention includes registering the vertical feature line in the vertical line database when the detected vertical feature line is not a line registered in a vertical line database, If not present, the method may include deleting from the vertical line database and measuring the azimuth and position coordinates of the vertical line when the vertical feature line exists after the photographing position change.

In addition, in one aspect of the present invention, the position estimation step according to the environment includes determining whether the ceiling feature point exists, correcting the physical coordinate position of the ceiling feature point if the ceiling feature point exists, and If the feature point does not exist, the method may include measuring the rotation change amount of the moving object from the vertical feature point, and correcting the position by measuring the odometer change amount and the moving object rotation change amount.

In addition, the present invention may further include a driving map making step of simultaneously recording the ceiling feature points and the wall vertical feature lines on the driving map.

In another aspect of the present invention, the method may further include controlling driving of the moving object using the estimated position information of the moving object.

According to the present invention, when the position of the moving object is estimated from the natural or artificial markings of the ceiling image obtained in the environment in which the moving object is located, and the ceiling image lacking the environmental change or the feature point of the moving object is obtained, based on the vertical feature line map on the wall. By correcting the position of the moving object, it is possible to more accurately perform the position estimation and the mapping of the moving object.

In addition, according to the present invention, the mapping range can be extended by simultaneously receiving not only the ceiling image but also the wall omnidirectional image.

In addition, according to the present invention, since it can be implemented by a single camera it is possible to simplify the manufacturing cost and manufacturing process cheap.

Hereinafter, a moving system and a method for recognizing a position of a moving object will be described in detail with reference to the accompanying drawings. However, the present invention is not limited to or limited by the embodiments. Like reference numerals in the drawings denote like elements.

1 is a view showing the configuration of a mobile system according to an embodiment of the present invention.

Referring to FIG. 1, the movement system 100 according to an embodiment of the present invention includes a ceiling / wall omnidirectional image input device 110 and a moving object driving control device 120, and includes a ceiling / wall omnidirectional image input device ( The driving position of the moving object is controlled by recognizing the position of the moving object using the ceiling / wall omnidirectional image input from 110.

The ceiling / wall omnidirectional image input device 110 includes a lens adapter 111 for integrating ceiling and wall light and a camera 115 for acquiring an input image, and includes an upper end of the movable body in a direction perpendicular to the bottom surface. Is installed on.

FIG. 2 is a diagram illustrating a structure of a ceiling / wall omnidirectional image input device shown in FIG. 1.

Referring to FIG. 2, the lens adapter 111 receives the ceiling and the wall omnidirectional image simultaneously with the camera 115, and has a convex reflection mirror 112 and a reflection mirror 112 having a hole in the center thereof. It consists of a lens support 114 for positioning the wide-angle lens 113 and the reflecting mirror 112 mounted in the center hole at a predetermined distance from the camera 115.

The lens adapter 111 mounts a convex reflection mirror 112 on the front of the camera 115 in a direction perpendicular to the floor, and reflects an image of the wall and the bottom surface 360 degrees around the center axis of the camera 115. It is provided to the camera 115. At this time, the convex reflection mirror 112 has a hole penetrating through the center, and the wide-angle lens 113 is mounted in the hole to receive a wide range of images incident from the ceiling through the wide-angle lens 113, the camera 115 To provide. For example, the block-type reflection mirror 112 may be hyperbolic, and the wide-angle lens 113 may enlarge the image recognition range of the ceiling / wall omnidirectional image input device 110 by using an aspherical wide-angle lens.

3 is a diagram illustrating an image recognition range.

Referring to FIG. 3, the ceiling / wall omnidirectional image input device 110 receives a ceiling / wall omnidirectional image for a ceiling lens recognition region and an omnidirectional lens recognition region. For example, the ceiling viewing angle is 120 degrees, and the wall viewing angle may have a viewing angle with respect to 360 degrees within 60 degrees in a direction facing the wall.

The camera 115 is composed of a condenser lens, an image capturing apparatus, and an image capturing apparatus which stores an image and transmits the image to the image processor in order to convert the light received by the lens adapter 111 into an image. For example, the camera 115 may use a camera capable of detecting near infrared rays, and may acquire both visible and near infrared band light.

The moving object driving control apparatus 120 includes a driving unit 121 for moving driving such as a wheel or a leg, a sensor unit 122 providing control information for preventing fall or collision, and a ceiling / wall omnidirectional image input device 110. And a moving object controller 123 for estimating the position of the moving object using the ceiling / wall omnidirectional image feature information captured from the image, and controlling the driving unit 121 using the estimated position information.

As described above, the moving system according to the present invention estimates the position of the moving object from the natural mark or the artificial mark of the ceiling image obtained in the environment in which the moving object is located, and acquires the ceiling image lacking the environmental change or the feature point of the moving object. By correcting the position of the moving object based on the line map, the position estimation and the mapping of the moving object can be performed more accurately.

4 is a flowchart illustrating an operation procedure of a method of estimating a moving object and generating a driving map according to an exemplary embodiment of the present invention.

1 to 4, in step S410, the ceiling / wall omnidirectional image input apparatus 110 simultaneously acquires a ceiling image and a wall omnidirectional image. The ceiling image is positioned in a circular ceiling image region with respect to the camera origin of the acquired image, and the wall omnidirectional image is positioned outside the ceiling image region in the circular wall image region. For example, in operation S410, the ceiling / wall omnidirectional image input device 110 may acquire an image of a ceiling viewing angle of 90 degrees, a wall viewing angle of 45 degrees, and a 360 degrees omnidirectional viewing angle as an image of one frame.

7 is a diagram illustrating an example of an acquired ceiling / wall omnidirectional image.

Referring to FIG. 7, in the acquired image, a ceiling region and a wall omnidirectional region exist on one image data in the outer region of the circle. The wall image is captured in a direction in which the height of the wall increases in an outward direction from the center of the image, and a vertical line of the wall is captured in a direction converging to the center point of the image. The inner region of the circle is a ceiling image area of 2.5 m in height from the moving object, and a ceiling image of 2.5 m in diameter is captured in the circular area. Vertical lines on the wall are picked up to increase outward from the image but increase in height toward the center of the image under the ceiling area. However, when the vertical line is close to the moving object, the vertical line is simultaneously imaged in the ceiling area and the wall area, and exists in the same orientation about the image center point.

In operation S420, the moving object controller 123 extracts an image feature from the obtained ceiling image and the wall omnidirectional image. That is, in step S420, the moving object controller 123 extracts a ceiling feature point from the obtained ceiling image, and extracts a wall vertical feature line from the wall omnidirectional image.

5 is a diagram illustrating an example of an operation of an image feature extraction step.

Referring to FIG. 5, in operation S510, the moving object controller 123 extracts a ceiling area from the obtained ceiling / wall omnidirectional image (S511), and corrects ceiling image distortion (S512). That is, in step S510, the moving object controller 123 extracts the ceiling image region from the lens distortion image generated by the wide angle lens 113, and corrects the wide angle lens distortion of the extracted ceiling image region. to be.

In operation S520, the moving object controller 123 extracts a ceiling feature point from the corrected ceiling image. That is, in operation S520, the moving object controller 123 extracts the ceiling feature point by an image processing algorithm that extracts contour and vertex information of the ceiling area from the corrected ceiling image. Since the ceiling feature point can observe only a certain area due to the limitation of the viewing angle of the ceiling area, if it is not observed, it is calculated by a dead reckoning algorithm, and the dead reckoning has a long moving distance due to the cumulative error of the sensor. The position error may increase due to the accumulation of ground sensor errors. On the other hand, in the case of the wall omnidirectional image, even if the distance of the moving object away from the wall, the wall size decreases but does not disappear from the field of view, so the limitation of the viewing distance of the wall surface can be solved by improving the resolution of the camera image sensor. When the estimation method is applied based on the wall omnidirectional image, the position error is greatly reduced because the cumulative error of the sensor converges within the azimuth error extracted from the wall omnidirectional image.

In operation S530, the moving object controller 123 detects a line component that converges to the image center from the obtained ceiling / wall omnidirectional image (S531), and whether the detected line component exists simultaneously in the ceiling image and the wall image. Judge. That is, step S530 is a vertical feature line detection step in which the moving object controller 123 detects a vertical feature line that converges to the image center from the obtained ceiling / wall image.

When the detected line component is present in the ceiling image and the wall image at the same time, in step S540, the moving object controller 123 determines whether the line is registered in the existing DB (S541), and in the case of the unregistered line, the vertical line It is registered in the DB (S542), and in the case of the registered line, it is determined whether the line exists even after the change of the shooting position (S543). If the line does not exist even after the change of the shooting position, the information is deleted from the vertical line DB (S544). . If a line exists even after the photographing position change, the azimuth and position coordinates of the vertical line are measured. That is, in step S540, the moving object controller 123 temporarily stores the vertical line information detected in the vertical feature line detection step S530 in the vertical line DB, and continuously tracks when the moving object moves so that the vertical line is changed even after a certain moving position is changed. When present, it is a vertical feature line verification step of recognizing a strong vertical feature line and storing it on the vertical line DB.

In operation S550, the moving object controller 123 generates moving object position feature data using the extracted ceiling feature point information and the verified vertical feature line information. That is, in step S550, the moving object controller 123 combines the extracted ceiling feature point information and the wall vertical feature line information together with moving object position feature data to collect position feature data of a set of moving objects from an input image of one frame. Create

In operation S430, the moving object controller 123 estimates the position of the moving object according to the environment by using the extracted image feature. That is, in step S430, the moving object controller 123 selects a location estimation method (S431), estimates a location using a ceiling feature point according to the selected location estimation method (S432), or uses a vertical wall line to determine a location. It is estimated (S433).

In step S440, the moving object controller 123 creates a driving map according to the estimated position of the moving object.

FIG. 6 is a diagram illustrating an example in which a moving object position estimating step and a traveling map preparation step according to an environment are specified.

Referring to FIG. 6, in step S430, the moving object controller 123 determines whether a ceiling feature point exists from the moving object position feature data (S611), and if the ceiling feature point exists, corrects the physical coordinate position of the ceiling feature point. (S612) If the ceiling feature point does not exist, the moving object rotation change amount is measured from the vertical feature line (S613), and the position by the estimation of the odometer change amount and the moving object rotation change amount is corrected (S614). That is, in step S430, the moving object controller 123 observes the feature points on the ceiling area and estimates the position of the moving object based on the ceiling feature points when the ceiling feature points exist, and does not have the ceiling feature points. In the environment where the feature point is lost due to the change, the position of the moving object can be estimated by tracking the azimuth change of the wall omnidirectional vertical feature line.

In operation S440, the moving object controller 123 determines whether a similar feature point is searched on the map DB (S615), and if it is determined that the similar feature point is not found, stores the new feature point and the vertical feature line on the map. If it is determined that the similar feature point is found (S616), the position of the moving object is estimated (S167) to generate a driving map.

In operation S450, the moving object controller 123 controls the driving of the moving object using the estimated position information of the moving object. In this case, the moving object controller 123 may provide the wheel encoder information when the position is corrected in step S614.

As described above, the method of estimating the position of the moving object according to the present invention estimates the position of the moving object from the natural or artificial markings on the ceiling image obtained in the environment in which the moving object is located, and acquires the ceiling image lacking the environmental change or the feature point of the moving object. By correcting the position of the moving object based on the vertical feature line map of the image, the position estimation and the mapping of the moving object can be performed more accurately.

As described above, the present invention has been described by way of limited embodiments and drawings, but the present invention is not limited to the above embodiments, and those skilled in the art to which the present invention pertains various modifications and variations from such descriptions. This is possible.

Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined not only by the claims below but also by the equivalents of the claims.

1 is a view showing the configuration of a mobile system according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating a structure of a ceiling / wall omnidirectional image input device shown in FIG. 1.

3 is a diagram illustrating an image recognition range.

4 is a flowchart illustrating an operation procedure of a method of estimating a moving object and generating a driving map according to an exemplary embodiment of the present invention.

5 is a diagram illustrating an example of an operation of an image feature extraction step.

FIG. 6 is a diagram illustrating an example in which a moving object position estimating step and a traveling map preparation step according to an environment are specified.

7 is a diagram illustrating an example of an acquired ceiling / wall omnidirectional image.

<Explanation of symbols for the main parts of the drawings>

100: moving system

110: ceiling / wall omnidirectional video input device

111: lens adapter

112: convex reflection mirror

113: wide angle lens

114: lens support

115: camera

120: moving object drive control device

121: drive unit

122: sensor unit

123: moving object control unit

Claims (10)

A driving unit for driving the moving body; A sensor unit providing control information for controlling the driving unit; A ceiling / wall omnidirectional image input device installed in a direction perpendicular to the floor to capture ceiling / wall omnidirectional images; And A moving control unit for estimating the position of the moving object using feature information of the ceiling / wall omnidirectional image captured by the ceiling / wall omnidirectional image input unit and controlling the driving unit using the estimated position information Moving system comprising a. The method of claim 1, The ceiling / wall omnidirectional image input device, A lens adapter composed of a convex reflection mirror with a hole in the center, a wide-angle lens mounted in the center hole of the reflection mirror, and a lens support for positioning the reflection mirror at a predetermined distance from the camera; And The lens adapter is mounted to the front, the camera to capture the ceiling / wall omnidirectional image Including, The mobile system is installed on the top of the movable body in a direction perpendicular to the bottom surface. An image acquisition step of simultaneously acquiring a ceiling and a wall omnidirectional image through a ceiling / wall omnidirectional image input means; An image feature extraction step of extracting feature points and feature lines from the obtained ceiling and wall omnidirectional images; And Position estimation step according to an environment for estimating the position of the moving object by selecting a position estimation method of the moving object according to the extracted feature Position estimation method of the moving body comprising a. The method of claim 3, The image feature extraction step, Detecting a vertical feature line converging at the center of the obtained ceiling image; Verifying a robust vertical feature line by tracking a vertical feature line according to a change in position of the moving object; Correcting distortion of the wide-angle lens in the ceiling area; Extracting ceiling feature points from the corrected ceiling image; And Generating moving object position feature data comprising the extracted wall vertical feature lines and the ceiling feature points Including, the position estimation method of the moving body. The method of claim 4, wherein The location estimation step according to the environment, If the ceiling feature points exist, the ceiling feature points are tracked to estimate the position of the moving object. And estimating the position of the moving object by tracking an azimuth change amount of the wall vertical feature line when the ceiling feature point does not exist. The method of claim 4, wherein Verifying the vertical feature line, Registering in the vertical line database if the detected vertical feature line is not a line registered in a vertical line database; If the vertical feature line does not exist after the change of a photographing position, deleting from the vertical line database; And Measuring the azimuth and position coordinates of the vertical line when the vertical feature line exists after a change in a photographing position; Comprising a moving position estimation method. The method of claim 5, The location estimation step according to the environment, Determining whether the ceiling feature point exists; Correcting a physical coordinate position of the ceiling feature point if the ceiling feature point exists; If the ceiling feature point does not exist, measuring a change amount of rotation of the moving object from a vertical feature point; And Correcting the position by measuring the odometer change amount and the moving object rotation change amount Position estimation method of the moving body comprising a. The method of claim 3, And a driving map making step of simultaneously recording the ceiling feature points and the wall vertical feature lines on the driving map. The method of claim 8, The driving map making step, Searching for similar feature points on the map database; And If the similar feature point is not found, storing a new feature point and a vertical feature line on a map Position estimation method of the moving body comprising a. The method of claim 3, And controlling driving of the movable body by using the estimated position information of the movable body.

Images