Autonomous flying drone monitoring underground power lines

The autonomous drone system addresses safety and efficiency issues in underground power conduit inspection by using a marker strip for stable flight and image capture, enhancing accuracy and reducing maintenance time.

KR103046735B1Active Publication Date: 2026-11-11E-ENGINEERING CO LTD
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
KR1020260151654
Authority / Receiving Office
KR · KR
Patent Type
Patents
Current Assignee / Owner
Filing Date
2026-08-13
Publication Date
2026-11-11
Estimated Expiration
2046-08-13

AI Technical Summary

Technical Problem

Conventional methods for inspecting underground power conduits face challenges such as safety risks for workers, low efficiency, and difficulty in maintaining stable drone flight due to low illumination, repetitive structures, and poor wireless communication, leading to inaccurate altitude and direction control.

Method used

An autonomous drone equipped with a marker camera and gimbal system uses a marker strip on the conduit ceiling to calculate altitude, left-right, and direction corrections, enabling stable flight and image capture, even in curved sections, and stores footage for post-analysis.

Benefits of technology

Enables safe, efficient, and accurate monitoring of underground power conduits by maintaining drone distance, direction, and location identification, improving maintenance efficiency and reducing accidents.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 112026099229250-PAT00019_ABST
    Figure 112026099229250-PAT00019_ABST
Patent Text Reader

Abstract

The present invention relates to an autonomous flight drone for monitoring underground power distribution lines, and more specifically, to an autonomous flight drone for monitoring underground power distribution lines that can autonomously fly inside an underground power conduit, capture surveillance images of the underground power conduit and underground power distribution lines, save the captured surveillance images, and then transmit them while landed at a docking station. The present invention relates to an autonomous flight drone capable of stably controlling the distance from the ceiling, left and right position, and direction of travel inside an underground power conduit by capturing a marker band marked on the ceiling of the underground power conduit and a centerline marked at the center of the width direction of the marker band using a marker camera, and generating altitude correction values, left and right correction values, and direction correction values ​​by analyzing the captured identification images.
Need to check novelty before this filing date? Find Prior Art

Description

Technology Field

[0001] The present invention relates to an autonomous flight drone for monitoring underground power distribution lines, and more specifically, to an autonomous flight drone for monitoring underground power distribution lines that can autonomously fly inside an underground power conduit, capture surveillance images of the underground power conduit and underground power distribution lines, save the captured surveillance images, and then transmit them while landed at a docking station. The present invention relates to an autonomous flight drone capable of stably controlling the distance from the ceiling, left and right position, and direction of travel inside an underground power conduit by capturing a marker band marked on the ceiling of the underground power conduit and a centerline marked at the center of the width direction of the marker band using a marker camera, and generating altitude correction values, left and right correction values, and direction correction values ​​by analyzing the captured identification images. Background Technology

[0002] Underground power conduits, where underground power distribution lines are laid, are underground facilities that house various cables such as distribution lines, communication lines, and control lines, as well as various devices such as cable support hangers, junction boxes, ventilation equipment, drainage equipment, lighting equipment, and fire detection equipment. Since these underground power conduits are major infrastructure facilities directly related to the stability of power supply, it is necessary to periodically inspect for abnormal conditions such as cable deterioration, overheating, leakage, flooding, condensation, signs of fire, damage to cable support structures, ingress of foreign matter, and equipment failure.

[0003] To this end, conventional methods mainly involved workers directly entering underground power tunnels to inspect them visually, or using equipment such as portable cameras, thermal imaging cameras, and gas detectors to check the internal conditions. However, since underground power tunnels are environments where passageways are narrow, lighting is low, and temperatures and humidity are high, as well as dust, leaks, and hazardous gases may be present, the method of workers directly entering to inspect poses a high risk of safety accidents and has low inspection efficiency.

[0004] Furthermore, underground power conduits feature a linear structure that extends long in the longitudinal direction, often with cables, hangers, junction boxes, and various auxiliary facilities arranged in a repetitive manner. Consequently, inspecting the entire section by physically moving through it consumes a significant amount of time, and the detection of abnormal conditions can vary depending on the inspector's proficiency or fatigue level. Particularly in sections where identical or similar structures appear repeatedly, it is difficult to pinpoint the exact location even if signs of anomalies are detected, which may result in additional time being required for subsequent maintenance work.

[0005] Conventional technologies intended to address this issue have considered methods involving deploying mobile inspection devices or unmanned vehicles into underground power conduits to capture video. However, since the floors of underground power conduits contain drainage channels, elevation differences, cables, obstacles, flooded areas, or work debris, devices moving along the floor suffer from reduced driving stability. Furthermore, the inspectable sections are limited by obstacles or flood conditions, making practical use difficult.

[0006] To address these issues, a technology for automatically patrolling underground power conduits using autonomous drones is being proposed. The advantages of using autonomous drones include reducing the frequency of workers directly entering underground power conduits, enabling relatively rapid filming of long sections of the conduit, and allowing filming of areas that are difficult for workers to access, such as near the ceiling, around cable hangers, or above cables.

[0007] However, unlike typical outdoor flight environments, the interior of underground power tunnels cannot utilize GPS, has very low illumination, and features a structure where identical cables and cable hangers are repeated over long distances, making it difficult to directly apply standard autonomous flight technology. Furthermore, due to the presence of various structures inside the tunnels—such as cables, cable hangers, junction boxes, wall protrusions, lighting equipment, and piping—it is difficult for drones to perform long-distance flights while maintaining a sufficient safety distance.

[0008] In addition, while typical underground power conduits are formed with a width and height of approximately 2 to 4 meters, thereby securing space for drone flight, considering the various structures and facilities installed near the ceiling, drones often need to fly long distances while maintaining a constant distance of several tens of centimeters from the ceiling. In this case, since even a change in height of only a few centimeters can significantly alter the distance from the ceiling, it is very important to continuously maintain a constant distance during long-distance autonomous flight.

[0009] However, when using general ultrasonic sensors, infrared sensors, optical distance sensors, LiDAR, or vision sensors, measurement errors may occur depending on the ceiling material or surface condition, lighting environment, reflectivity, dust, condensation, etc., making it difficult to consistently maintain high precision during long-distance flights. In particular, since the interior of underground power tunnels has low illumination and may be affected by reflections and shadows caused by metal cables or cable hangers, relying solely on simple distance measurements or general image feature points limits the ability to stably control the drone's altitude, left-right position, and direction of travel.

[0010] Furthermore, the roll or pitch of an autonomous drone may change during the process of performing speed changes, acceleration / deceleration, attitude control, direction changes, or flight through curves. When such attitude changes occur, the attitude of the camera capturing the ceiling may also tilt, resulting in distortion of images of reference objects or flight targets formed on the ceiling. This image distortion can cause errors in pixel count, center position, or tilt judgment, thereby reducing the reliability of altitude and path control.

[0011] Furthermore, the interior of underground power conduits may include curved sections as well as straight sections. In curved sections, it is difficult to accurately determine whether a drone has entered the curve and its direction of travel if it relies solely on recognizing obstacles or walls ahead. Moreover, during long-distance flights, the drone may lose its normal direction of travel if disturbances, sudden yaw turns, or control errors occur. In such cases, the drone may fly in the wrong direction or approach cable hangers or walls, increasing the risk of collision.

[0012] Furthermore, since the interiors of underground power conduits mostly consist of repetitively arranged identical cables and cable hangers, it is difficult to identify the specific location from the recorded surveillance footage alone. Consequently, even if abnormal signs are detected in the footage, it is difficult for workers to quickly locate the specific site, which can reduce the efficiency of maintenance work.

[0013] Furthermore, the interior of underground power conduits is very poor for establishing a wireless communication environment. Wireless signals can be attenuated or reflected by power cables, metal structures, and concrete walls, and if the underground power conduit is formed as a long, enclosed structure, there is a problem in that it is difficult to perform stable real-time communication with a remote monitoring server while the drone is flying.

[0014] Therefore, for the unmanned monitoring of underground power conduits and distribution lines, autonomous flight drone technology for monitoring underground distribution lines is required that can reliably acquire surveillance images inside the conduit without direct worker entry, precisely maintain distance from the ceiling, left-right position, and direction of travel in enclosed spaces where GPS use is difficult, and perform stable autonomous flight even in curved sections and sections with repetitive structures. Prior art literature

[0015] Korean Registered Patent No. 10-2487554 (2023-01-06) System for Detecting Internal Damage in Utility Tunnels Using Autonomous Mobile Vehicles Korean Published Patent No. 10-2019-0091714 (2019-08-07) Autonomous Exploration Method for Underground Utility Tunnels Using Unmanned Aerial Vehicles Korean Registered Patent No. 10-2710776 (2024-09-23) Power Tunnel Inspection Device Korean Registered Patent No. 10-2019049 (2019-09-02) Mobile Robot for Maintenance of Underground Structures The problem to be solved

[0016] The present invention, devised to solve the aforementioned problems, aims to provide an autonomous flying drone for monitoring underground power lines that enables periodic acquisition of surveillance images of underground power conduits and underground power lines without a worker having to directly enter the underground power conduit.

[0017] Another objective of the present invention is to enable an autonomous drone to fly stably along a predetermined inspection path even inside an underground power tunnel where it is difficult to use a Global Positioning System (GPS), lighting is low, and cables, cable hangers, junction boxes, and various auxiliary facilities are repeatedly arranged.

[0018] Another objective of the present invention is to enable an autonomous flight drone to rapidly determine an increase or decrease in distance from the ceiling and perform ascent or descent control accordingly in an environment where it must fly while maintaining a short distance of about tens of centimeters from the ceiling of an underground power conduit.

[0019] Another objective of the present invention is to stably maintain the distance (H) from the ceiling without relying solely on a separate precision distance sensor by photographing a marker strip of a certain width (D) marked on the ceiling of an underground power conduit with a marker camera and calculating the difference between a reference pixel (P0) and a measured pixel (P1) using the number of pixels in the width direction of the marker strip captured in the identification image.

[0020] Another objective of the present invention is to determine the left-right direction deviation and forward direction error of an autonomous flying drone using a center line (C0) marked at the center of the width direction of a marker strip, and accordingly, to perform left-right direction horizontal movement control and horizontal angle (yaw) control.

[0021] Another objective of the present invention is to enable an autonomous flight drone to fly along the space below the center line (C0) by utilizing the relative left-right position between a center point (Cp1) located at the exact center of the identification image and a center line midpoint (Cp2) that bisects the image portion of the center line (C0) captured in the identification image in the longitudinal direction.

[0022] Another objective of the present invention is to enable the correction of the horizontal angle (yaw) so that the direction of travel of an autonomous flight drone remains parallel to the longitudinal direction of the center line (C0) by utilizing the difference in the intersection angle or slope between the image portion of the center line (C0) captured in the identification image and the center vertical line (C1) of the identification image.

[0023] Another objective of the present invention is to reduce distortion in identification images and improve the reliability of altitude control, left-right control, and direction control by controlling the marker gimbal so that the optical axis of the marker camera lens maintains a vertical direction toward the ceiling of the underground power conduit, even if roll or pitch occurs during the process of the autonomous flight drone performing speed changes, acceleration / deceleration, direction changes, or attitude control.

[0024] Another objective of the present invention is to determine the entry or exit of a curved section using curve markers displayed at the starting and ending positions of the curved section, even when the underground power conduit includes a curved section, and to correct the direction of travel of an autonomous flight drone to correspond to the radius of curvature (R) and the direction of bending.

[0025] Another objective of the present invention is to enable an autonomous drone to stably pass through a curved section without separate complex terrain recognition by distinguishing the starting and ending positions of the curved section using the identification order of curve markers.

[0026] Another objective of the present invention is to enable stable control of the distance from the ceiling while utilizing a wide angle of view by calculating a reference pixel (P0) by reflecting the projection characteristics of the fisheye lens, even when the marker camera includes a fisheye lens with an equidistant projection method.

[0027] Another objective of the present invention is to generate path missing information when a marker band or centerline (C0) is not identified in an identification image, and to enable the autonomous flight drone to return to a docking station by following the previous flight path in reverse or changing its direction of travel by 180 degrees, thereby enabling the autonomous flight drone to be safely recovered even in a situation where the path is lost.

[0028] Another objective of the present invention is to divide the marker band into a first band area located to the left of the center line (C0) and a second band area located to the right, and to determine whether the autonomous flight drone is in an inspection flight state or a return flight state by using the difference in color, brightness, or saturation between the first band area and the second band area.

[0029] Another objective of the present invention is to enable the rapid and accurate identification of the location where an anomaly is detected in a surveillance video by storing the location identification ID or distance traveled included in the location marker as metadata of the surveillance video, even when identical or similar cables and cable hangers are repeatedly arranged inside an underground power conduit.

[0030] Another objective of the present invention is to enable an autonomous drone to store surveillance footage captured during flight in a storage unit and transmit the footage to a docking station via a local area network after landing, even when real-time communication with a remote monitoring server is difficult during autonomous flight due to poor wireless communication environments inside underground power conduits.

[0031] Another objective of the present invention is to enable stable image acquisition using only relatively simple gimbal control that maintains a vertical upward direction relative to the direction of gravity, rather than high-difficulty gimbal control that precisely tracks the local normal direction of the ceiling surface, by taking into account the fact that typical underground power conduits may have a gentle slope in the direction of travel but mostly remain horizontal in the width direction.

[0032] The technical problems that the present invention aims to solve are not limited to those mentioned above, and other unmentioned technical problems will be clearly understood by those skilled in the art to which the present invention belongs from the description below. means of solving the problem

[0033] To solve the aforementioned problems, the present invention relates to an autonomous flying drone for monitoring underground power distribution lines, capable of autonomously flying inside an underground power conduit and capturing surveillance images of the underground power conduit and underground power distribution lines, wherein the ceiling of the underground power conduit is provided with a marker strip formed in the shape of a strip having a certain width (D) and having a color contrasting with the color of the ceiling surface of the underground power conduit, extending from the inspection start point to the inspection end point of the underground power conduit, wherein the marker strip is provided with a center line (C0) located at the center in the width direction of the marker strip and extending along the length direction of the marker strip, wherein the center line (C0) has a color, brightness, or saturation distinct from the color of the marker strip, and the autonomous flying drone comprises a flight control unit, a surveillance camera, a marker camera, a marker gimbal, a control calculation unit, a storage unit, and a communication unit, wherein the surveillance camera captures surveillance images of the underground power conduit and underground power distribution lines while the autonomous flying drone is flying and stores them in the storage unit The marker camera captures an identification image of the marker band and the centerline (C0) while the autonomous flying drone is flying and transmits it to the control computing unit in real time; the marker gimbal controls the attitude of the marker camera so that the optical axis of the marker camera's lens is vertical toward the ceiling of the underground power conduit; the storage unit stores a reference pixel (P0), which is the number of pixels in the width direction of the marker band captured when the marker camera captures the identification image at a certain distance (H) from the ceiling of the underground power conduit; the communication unit can connect to the docking station via a short-range communication network and transmit the surveillance image stored in the storage unit to the docking station when the autonomous flying drone lands at a docking station installed in the underground power conduit; and the control computing unit analyzes the marker band and the centerline (C0) captured in the identification image to obtain an altitude correction value (Hc),After generating a left-right correction value (Lc) and a direction correction value (Dc), they are transmitted to the flight control unit, and the flight control unit controls the autonomous flight drone to take off from the docking station, fly autonomously along the space below the marker strip from the inspection start point to the inspection end point, and then return to the docking station, wherein: - altitude control is performed using the altitude correction value (Hc) so that the autonomous flight drone can fly forward while maintaining the separation distance (H) from the ceiling of the underground power conduit; - horizontal movement control is performed using the left-right correction value (Lc) so that the autonomous flight drone can fly forward along the space below the centerline (C0); - horizontal angle (yaw) control is performed using the direction correction value (Dc) so that the autonomous flight drone can fly forward while maintaining its direction of travel parallel to the length direction of the centerline (C0); and the reference pixel (P0) is the width (D) of the marker strip, the An autonomous flight monitoring an underground power distribution line is characterized by being calculated using the separation distance (H) and the specifications of the marker camera, and the value calculated by determining the number of pixels in the width direction of the marker band image captured in the identification image is the actual pixel (P1); the altitude correction value (Hc) is generated based on the difference between the actual pixel (P1) and the reference pixel (P0); the left-right correction value (Lc) is generated based on the left-right relative position between the center point (Cp1) located at the exact center of the identification image and the center line midpoint (Cp2); the direction correction value (Dc) is generated based on the difference in intersection angle or slope between the center line (C0) and the center vertical line (C1) captured in the identification image; the center line midpoint (Cp2) is the point that bisects the image portion of the center line (C0) captured in the identification image in the length direction; and the center vertical line (C1) is a line segment connecting the upper side midpoint and the lower side midpoint of the identification image. It is advisable to do this with a drone.

[0034] In addition to the features described above, the present invention further comprises, in addition to the above-described features, the specifications of the marker camera including a focal length (f), a horizontal length (W) of the image sensor, and a total number of horizontal pixels (Pw) of the image sensor, wherein the reference pixel (P0) is calculated according to the following formula using the width (D) of the marker band, the separation distance (H), the focal length (f), the horizontal length (W) of the image sensor, and the total number of horizontal pixels (Pw) of the image sensor.

[0035]

[0036] It is also preferable to have an autonomous flight drone for monitoring an underground power distribution line, wherein the altitude correction value (Hc) is a value obtained by subtracting the actual pixel (P1) from the reference pixel (P0) (P0-P1), or a value proportional to the subtracted value (P0-P1), and the flight control unit: - maintains the current altitude of the autonomous flight drone when the altitude correction value (Hc) is within a predetermined allowable range; - performs descent control so that the autonomous flight drone moves away from the ceiling of the underground power conduit when the altitude correction value (Hc) is a negative value outside the allowable range; - performs ascent control so that the autonomous flight drone moves closer to the ceiling of the underground power conduit when the altitude correction value (Hc) is a positive value outside the allowable range; and - increases the amount of control for the ascent control or the descent control as the absolute value of the altitude correction value (Hc) increases.

[0037] In addition to the features described above, the control calculation unit calculates the number of left-right relative pixels (P2) between the center point (Cp1) and the center line midpoint (Cp2), and calculates the left-right deviation distance (d1) of the autonomous flight drone according to the following formula using the width (D) of the marker band, the actual pixel (P1), and the number of left-right relative pixels (P2).

[0038]

[0039] It is also preferable to have an autonomous flight drone for monitoring an underground power distribution line, characterized by generating a left-right correction value (Lc) including a deviation direction indicating whether the left-right deviation distance (d1) and the centerline midpoint (Cp2) are located to the left or right relative to the center point (Cp1), and transmitting this to the flight control unit, and the flight control unit performing left-right horizontal movement control of the autonomous flight drone using the left-right correction value (Lc) so that the centerline midpoint (Cp2) in the identification image coincides with the center point (Cp1).

[0040] In addition to the features described above, the present invention monitors an underground power distribution line, wherein the direction correction value (Dc) is the intersection angle (θ) formed by the image portion of the center line (C0) captured in the identification image with respect to the center vertical line (C1), or a value proportional to the intersection angle (θ), and has a positive value when the image portion of the center line (C0) is tilted clockwise and a negative value when it is tilted counterclockwise; the flight control unit maintains the current heading angle of the autonomous flight drone when the direction correction value (Dc) is within a predetermined allowable range, and performs horizontal angle (yaw) control for the autonomous flight drone so that the intersection angle (θ) approaches 0 when the direction correction value (Dc) deviates from the allowable range, and increases the amount of control for the horizontal angle (yaw) control as the absolute value of the direction correction value (Dc) increases, and the control calculation unit calculates the intersection angle (θ) as shown in the following equation. It is also desirable to use autonomous drones.

[0041]

[0042] (x, y are relative coordinates for an arbitrary point other than the intersection point in the image portion of the center line (C0), where the intersection point of the image portion of the center line (C0) and the center vertical line (C1) is set as the coordinate origin (0,0).)

[0043] In addition to the features described above, the present invention further comprises, wherein the underground power conduit includes a curved section, and at each of the starting and ending positions of the curved section, a curved marker is displayed that is placed within the area of ​​the marker band, and each of the curved markers includes an identification ID, a radius of curvature (R) of the curved section, and a bending direction when entering the curved section based on the corresponding curved marker; the control calculation unit, when the curved marker is identified in the identification image, generates curved section entry information or curved section end information using the identification ID, the radius of curvature (R), and the bending direction included in the curved marker; and the flight control unit, when receiving the curved section entry information, performs horizontal angle (yaw) control based on the direction correction value (Dc) so that the tangential direction or local length direction of the portion of the image part of the centerline (C0) located in front of the autonomous flight drone's direction of travel remains parallel to the center vertical line (C1), while [regarding] the radius of curvature (R) and the bending direction It is also preferable to have an autonomous flight drone for monitoring an underground power distribution line, characterized in that the flight control unit corrects the direction of travel of the autonomous flight drone to correspond to the curve section, and when the flight control unit receives information on the end of the curve section, it terminates the correction of the direction of travel corresponding to the curve section and controls the drone to fly forward along the length direction of the centerline (C0).

[0044] In addition to the features described above, the present invention is also preferably an autonomous flying drone for monitoring an underground power distribution line, wherein the control calculation unit: - counts the identification order of the curve markers when the curve markers are sequentially identified in the identification image while the autonomous flying drone is flying from the inspection start point to the inspection end point; - when the curve marker with an odd identification order is identified, determines the curve marker as a curve section start marker displayed at the starting point of the curve section; - when the curve marker with an even identification order is identified, determines the curve marker as a curve section end marker displayed at the end point of the curve section; - when the curve section start marker is identified, generates curve section entry information including the identification ID, the radius of curvature (R), and the bending direction; and - when the curve section end marker is identified, generates curve section end information for terminating the correction of the direction of travel corresponding to the curve section.

[0045] In addition to the features described above, the present invention further comprises, wherein the marker camera includes an equal-distance projection fisheye lens, and the specifications of the marker camera include the focal length (f) of the fisheye lens, the horizontal length (W) of the image sensor, and the total number of horizontal pixels (Pw) of the image sensor, and the reference pixel (P0) is calculated according to the following formula using the width (D) of the marker band, the separation distance (H), the focal length (f) of the fisheye lens, the horizontal length (W) of the image sensor, and the total number of horizontal pixels (Pw) of the image sensor.

[0046]

[0047] It is also preferable to have an autonomous flight drone for monitoring an underground power distribution line, wherein the control calculation unit calculates the number of pixels in the width direction of the marker band image captured in the identification image as the actual pixel (P1), and the altitude correction value (Hc) is a value obtained by subtracting the actual pixel (P1) from the reference pixel (P0) (P0-P1) or a value proportional to the subtracted value (P0-P1), and the flight control unit performs altitude control using the altitude correction value (Hc) so that the autonomous flight drone maintains the separation distance (H) from the ceiling of the underground power conduit.

[0048] In addition to the features described above, the present invention is also preferably an autonomous flying drone for monitoring an underground power distribution line, wherein the control calculation unit generates path absence information and transmits it to the flight control unit when the marker band or the centerline (C0) is not identified in the identification image, and the flight control unit, upon receiving the path absence information, controls the autonomous flying drone to follow the previous flight path in reverse or change its direction of travel by 180 degrees to return to the docking station, the storage unit stores the path absence information, the time of occurrence of the path absence information, the surveillance image at the time the path absence information occurred, or the identification image at the time the path absence information occurred, and the communication unit transmits the path absence information stored in the storage unit to the docking station when the autonomous flying drone lands at the docking station.

[0049] In addition to the features described above, the present invention further comprises, with respect to the direction from the inspection start point to the inspection end point, a first band area located to the left of the center line (C0) and a second band area located to the right of the center line (C0), wherein the first band area and the second band area have distinct colors, brightness, or saturation, and the storage unit stores the normal arrangement relationship of the first band area, the center line (C0), and the second band area that must be identified in the identification image when the autonomous flight drone flies from the inspection start point to the inspection end point, and the control calculation unit: - compares the current arrangement relationship of the first band area, the center line (C0), and the second band area identified in the identification image with the normal arrangement relationship; - if the current arrangement relationship matches the normal arrangement relationship, determines that the autonomous flight drone is in an inspection flight state heading from the inspection start point to the inspection end point; and - the current arrangement relationship It is also desirable to use an autonomous flight drone for monitoring an underground power distribution line, characterized by determining that the autonomous flight drone is in a return flight state heading from the inspection end point to the inspection start point when it is identified as being reversed left and right from the normal arrangement relationship mentioned above.

[0050] In addition to the features described above, the present invention is also preferably an autonomous flying drone for monitoring an underground power distribution line, wherein the marker strip further displays position markers at predetermined intervals from the inspection start point to the inspection end point, and the position markers include a position identification ID for identifying the location where the position markers are displayed within the underground power conduit or a distance traveled from the inspection start point to the location where the position markers are displayed, and the control calculation unit extracts the position identification ID or the distance traveled included in the position markers when the position markers are identified in the identification image, and the storage unit stores the position identification ID or the distance traveled as metadata of the surveillance image together with the surveillance image captured by the surveillance camera. Effects of the invention

[0051] According to the present invention, an autonomous drone can autonomously fly inside an underground power conduit and capture surveillance images of the underground power conduit and underground power distribution lines without a worker having to directly enter the underground power conduit. Accordingly, this has the effect of enabling safer and more efficient inspection of the condition of underground power distribution lines, even in environments where direct inspection by a worker is dangerous or inefficient due to high temperature, high humidity, leakage, flooding, condensation, harmful gases, or low light conditions.

[0052] In addition, according to the present invention, a marker strip having a certain width (D) is displayed on the ceiling of an underground power conduit, and an autonomous flying drone can calculate the difference between a reference pixel (P0) and a measured pixel (P1) by using the number of pixels in the width direction of the marker strip captured in the identification image. Accordingly, the autonomous flying drone can determine whether the distance (H) from the ceiling has increased or decreased based on the image without relying solely on a separate precision distance sensor, and has the effect of performing stable altitude control by reflecting the result of the determination in the ascent or descent control.

[0053] In addition, according to the present invention, since a center line (C0) is marked at the center of the width direction of the marker strip, the autonomous flight drone can control altitude using the width (D) of the marker strip, while simultaneously controlling the left and right position and direction of travel using the center line (C0). Accordingly, even in environments that are narrow and contain cables, cable hangers, wall structures, etc., such as inside an underground power conduit, the autonomous flight drone can fly stably along the space below the center line (C0).

[0054] In addition, according to the present invention, a left-right correction value (Lc) can be generated based on the left-right relative position between a center point (Cp1) located at the center of the identification image and a center line midpoint (Cp2). In particular, since the number of left-right relative pixels (P2) between the center point (Cp1) and the center line midpoint (Cp2) can be converted into a left-right deviation distance (d1) using the actual width (D) of the marker band and the actual pixels (P1), there is an effect of being able to perform left-right horizontal movement control that reflects the degree of deviation in actual space rather than simply comparing left-right positions on the image.

[0055] In addition, according to the present invention, a direction correction value (Dc) can be generated based on the difference in intersection angle or slope between the image portion of the center line (C0) captured in the identification image and the center vertical line (C1). Accordingly, even if the direction of travel deviates from the longitudinal direction of the center line (C0), the autonomous flight drone can correct the horizontal angle (yaw) to fly forward in a state parallel to the longitudinal direction of the center line (C0), and has the effect of reducing the error in the direction of travel that may accumulate during long-distance flight.

[0056] In addition, according to the present invention, the marker gimbal controls the lens optical axis of the marker camera to be vertical toward the ceiling of the underground power conduit. Accordingly, even if roll or pitch occurs during the process of the autonomous flight drone performing speed changes, acceleration / deceleration, direction changes, or attitude control, distortion of the identification image of the marker band and centerline (C0) can be reduced. As a result, the accuracy of calculating the number of pixels in the width direction of the marker band, the centerline center point (Cp2), the center vertical line (C1), and the intersection angle (θ) is improved, thereby increasing the reliability of altitude control, left / right control, and direction control.

[0057] In addition, the present invention takes into account that a general underground power conduit may have a gentle slope in the direction of travel but maintains a horizontal position in the width direction, and that the present invention performs altitude control, left-right position control, and direction of travel control using the number of pixels in the width direction of the marker strip (200) and the image position of the center line (C0). This allows for stable identification image acquisition with only a relatively simple gimbal control that maintains a vertical upward direction relative to the direction of gravity, rather than high-difficulty gimbal control that precisely follows the local normal direction of the ceiling surface.

[0058] In addition, according to the present invention, a reference pixel (P0) can be calculated using the width (D) of the marker band, the spacing (H), the focal length (f), the horizontal length (W) of the image sensor, and the total number of horizontal pixels (Pw) of the image sensor. Accordingly, a reference pixel (P0) reflecting the specifications of the marker camera and the target spacing (H) can be pre-set, and an altitude correction value (Hc) can be stably generated by comparing the actual pixel (P1) captured during flight with the reference pixel (P0).

[0059] In addition, according to the present invention, even when the marker camera includes a fisheye lens with an equidistant projection method, a reference pixel (P0) can be calculated by reflecting the projection characteristics of the fisheye lens. Accordingly, while capturing a marker band using a wide angle of view, an altitude correction value (Hc) can be generated by reflecting the image characteristics of the fisheye lens, which are different from general projection methods. Therefore, the accuracy of altitude control can be maintained even in environments where space is narrow and close-range shooting is required, such as inside an underground power conduit.

[0060] In addition, according to the present invention, when an underground power conduit includes a curved section, curve markers displayed at the starting and ending positions of the curved section can be used to generate information for entering or ending the curved section. Accordingly, an autonomous drone can pass through the curved section while correcting its direction of travel to correspond to the radius of curvature (R) and the direction of bending, and has the effect of performing stable autonomous flight based on the centerline (C0) not only in straight sections but also in curved sections.

[0061] In addition, according to the present invention, the identification order of curve markers is counted to determine that curve markers in odd order are start markers of a curved section and that curve markers in even order are end markers of a curved section. Accordingly, even without assigning separate, complex start and end distinction indicators to each curve marker, an autonomous flying drone can distinguish the entry and exit of a curved section and stably perform the initiation and termination of direction correction in the curved section.

[0062] In addition, according to the present invention, when a marker band or centerline (C0) is not identified in an identification image, path missing information is generated, and the autonomous flight drone can be controlled to follow the previous flight path in reverse or change its direction of travel by 180 degrees to return to the docking station. Accordingly, even if the flight path is lost due to marker recognition failure, poor lighting, obstacles, contamination, disturbances, or control errors, the autonomous flight drone can be safely recovered.

[0063] In addition, according to the present invention, the marker band includes a first band area and a second band area based on the center line (C0), and since the first band area and the second band area have colors, brightness, or saturation that are distinguishable from each other, the autonomous flight drone can determine the inspection flight state or the return flight state by comparing the current arrangement relationship in the identification image with the normal arrangement relationship. Accordingly, even if the autonomous flight drone recognizes the direction of travel in the opposite direction or the direction is reversed due to disturbances, it has the effect of identifying the flight state and performing a stable inspection flight or return flight.

[0064] In addition, according to the present invention, a location marker is displayed on a marker band, and a location identification ID or distance traveled included in the location marker can be stored as metadata of the surveillance video. Accordingly, when abnormal signs such as cable damage, overheating, leakage, flooding, cable hanger damage, or equipment failure are detected in the surveillance video, the location where the abnormal sign occurred can be identified quickly and accurately, and the efficiency of subsequent maintenance work can be improved.

[0065] Furthermore, according to the present invention, an autonomous drone can store surveillance footage in a storage unit while flying inside an underground power conduit and transmit the stored surveillance footage to the docking station via a local area network when it lands at a docking station. Accordingly, even when the wireless communication environment inside the underground power conduit is poor and real-time communication with a remote monitoring server is difficult during flight, there is an effect of stably collecting surveillance footage and providing it for post-analysis.

[0066] Furthermore, according to the present invention, since surveillance video can be transmitted to a surveillance server via a docking station, the surveillance server can analyze the received surveillance video to diagnose abnormalities in underground power conduits and underground power distribution lines. Accordingly, unmanned filming by an autonomous drone can be combined with remote analysis, and there is an effect of contributing to the prevention of accidents such as power outages, fires, or equipment damage by detecting abnormal conditions in underground power distribution lines at an early stage. Brief explanation of the drawing

[0067] FIG. 1 is a schematic diagram showing the state of an autonomous flying drone according to one embodiment of the present invention flying inside an underground power conduit based on a marker band and a centerline. FIG. 2 is a block diagram showing the configuration of an autonomous flight drone, a marker strip, a docking station, and a monitoring server according to one embodiment of the present invention. FIG. 3 is a diagram illustrating the principle of a marker camera according to an embodiment of the present invention capturing a marker band and a center line marked on the ceiling of an underground power conduit and calculating the number of pixels in the width direction of the marker band. FIG. 4 is a diagram illustrating the principle of controlling the altitude of an autonomous flight drone based on the difference between a reference pixel and a measured pixel according to an embodiment of the present invention. FIG. 5 is a diagram illustrating the principle of calculating left-right correction values ​​and direction correction values ​​using a center point, a centerline center point, a center vertical line, a number of relative pixels in the left-right direction, and an intersection angle in an identification image according to an embodiment of the present invention. FIG. 6 is a diagram illustrating the principle of controlling the flight of an autonomous drone in a curved section using a marker band, a centerline, and a curve marker marked on a curved section according to an embodiment of the present invention. FIG. 7 is a diagram illustrating the principle of calculating the number of pixels in the width direction of a marker band and performing altitude control using a marker camera including a fisheye lens of an equidistant projection method according to an embodiment of the present invention. Specific details for implementing the invention

[0068] The present invention will be described in detail below so that the aforementioned objectives and features may be clearly understood, and accordingly, a person skilled in the art to which the present invention pertains will be able to easily implement the technical concept of the present invention. Furthermore, in describing the present invention, regarding known technologies related to the present invention that are already well known in the technical field, if it is determined that a detailed description of such known technology might unnecessarily obscure the essence of the present invention, such detailed description will be omitted.

[0069] Furthermore, while the terms used in this invention have been selected to be as widely used as possible, there are also terms arbitrarily chosen by the applicant in specific cases; since the meanings of such terms are described in detail in the relevant description of the invention, it should be noted that the invention should be understood based on the meaning of the terms rather than their mere names. The terms used in the description of the embodiments are used merely to explain specific embodiments and are not intended to limit the embodiments. Unless the context clearly indicates otherwise, singular expressions include plural expressions.

[0070] Embodiments may be modified in various forms and may have various additional embodiments, and specific embodiments are shown in the drawings and described in detail. However, this is not intended to limit the embodiments to specific forms, and it should be understood that they include all modifications, equivalents, or substitutions that fall within the spirit and scope of the embodiments.

[0071] In the description of various embodiments, expressions such as “first,” “second,” “first,” or “second” may modify various components of the embodiments, but do not limit said components. For example, said expressions do not limit the order and / or importance of said components. Such expressions may be used to distinguish one component from another.

[0072] The present invention will be described below with reference to the attached drawings. First, a basic embodiment of the present invention will be examined. FIG. 1 is a schematic diagram showing the state in which an autonomous flying drone according to the present invention flies inside an underground power conduit based on a marker band and a centerline, and FIG. 2 is a block diagram showing the configuration of an autonomous flying drone, a marker band, a docking station, and a monitoring server according to the present invention.

[0073] Referring to FIG. 1, the underground power conduit (10) may include a ceiling (11), a floor (12), a cable hanger (13), and an underground power distribution line (14). The underground power distribution line (14) may be arranged along the longitudinal direction of the underground power conduit (10) while being supported by the cable hanger (13). An autonomous flying drone (100) may autonomously fly inside the underground power conduit (10) and capture surveillance images of the underground power conduit (10) and the underground power distribution line (14).

[0074] A marker strip (200) may be displayed on the ceiling (11) of the underground power conduit (10). The marker strip (200) may have a color that contrasts with the color of the ceiling (11) and may be formed in the shape of a strip with a constant width (D). The marker strip (200) may be displayed on the ceiling (11) so as to extend from the inspection start point to the inspection end point of the underground power conduit (10). Accordingly, the autonomous flying drone (100) can recognize a flight path based on the marker strip (200) inside the underground power conduit (10).

[0075] A center line (C0) may be displayed on the marker strip (200). The center line (C0) is located at the center of the width direction of the marker strip (200) and may extend along the length direction of the marker strip (200). The center line (C0) may have a color, brightness, or saturation that is distinct from the color of the marker strip (200). For example, if the marker strip (200) is formed in a bright color, the center line (C0) may be formed in a relatively dark color, and conversely, if the marker strip (200) is formed in a dark color, the center line (C0) may be formed in a relatively bright color. Accordingly, the autonomous flying drone (100) can distinguish and recognize the marker strip (200) and the center line (C0) from each other in the identification image.

[0076] Referring to FIG. 2, the autonomous flight drone (100) may include a flight control unit (110), a surveillance camera (120), a marker camera (130), a marker gimbal (131), a control calculation unit (140), a storage unit (150), and a communication unit (160). Additionally, the autonomous flight drone (100) may further include a sensor unit (170) according to an embodiment. The sensor unit (170) may include an inertial sensor, an accelerometer, a gyroscope, a barometric pressure sensor, an obstacle detection sensor, or a combination thereof, and the flight control unit (110) may control the attitude, altitude, or flight state of the autonomous flight drone (100) by using the detection value of the sensor unit (170) in an auxiliary manner.

[0077] A surveillance camera (120) can capture surveillance images of an underground power conduit (10) or an underground power distribution line (14) while the autonomous flying drone (100) is flying. The surveillance camera (120) may be composed of a general video camera, a thermal imaging camera, an infrared camera, or a combination thereof. The surveillance images captured by the surveillance camera (120) may be stored in a storage unit (150). The surveillance images may include an underground power distribution line (14), a cable hanger (13), a connection part, a ceiling (11), a wall surface, a floor (12), drainage facilities, lighting facilities, or other auxiliary facilities.

[0078] The marker camera (130) can capture identification images of the marker band (200) and the centerline (C0) while the autonomous flying drone (100) is flying. The marker camera (130) may be provided separately from the surveillance camera (120), and may be a camera that acquires identification images to control the altitude, left and right position, and direction of travel of the autonomous flying drone (100), rather than a camera for capturing surveillance images. The identification images captured by the marker camera (130) can be transmitted in real time to the control calculation unit (140).

[0079] The marker gimbal (131) can control the attitude of the marker camera (130). Specifically, the marker gimbal (131) can control the attitude of the marker camera (130) so that the lens optical axis of the marker camera (130) is in a vertical direction toward the ceiling (11) of the underground power conduit (10). Even if roll or pitch of the aircraft occurs during the process of the autonomous flying drone (100) moving forward, decelerating, accelerating, changing direction, or controlling attitude, the marker gimbal (131) can correct so that the lens optical axis of the marker camera (130) is maintained in a direction toward the ceiling (11). Accordingly, the number of pixels in the width direction of the marker band (200), the position of the center line (C0), and the inclination of the center line (C0) can be detected relatively stably in the identification image.

[0080] A reference pixel (P0) may be stored in the storage unit (150). The reference pixel (P0) may be the number of pixels in the width direction of the marker strip (200) captured in the identification image when the marker camera (130) captures the identification image at a certain distance (H) from the ceiling (11) of the underground power conduit (10). The reference pixel (P0) may be calculated in advance using the actual width (D) of the marker strip (200), the distance (H), and the specifications of the marker camera (130).

[0081] The control calculation unit (140) can generate an altitude correction value (Hc), a left / right correction value (Lc), and a direction correction value (Dc) by analyzing the identification image received from the marker camera (130). The control calculation unit (140) can calculate the number of pixels in the width direction of the marker band (200) in the identification image as the actual pixels (P1) and generate an altitude correction value (Hc) based on the difference between the actual pixels (P1) and the reference pixels (P0). If the actual pixels (P1) become larger than the reference pixels (P0), the autonomous flying drone (100) may be determined to be close to the ceiling (11), and if the actual pixels (P1) become smaller than the reference pixels (P0), the autonomous flying drone (100) may be determined to be far from the ceiling (11).

[0082] Additionally, the control calculation unit (140) can generate a left-right correction value (Lc) based on the left-right relative position between the center point (Cp1) located at the exact center of the identification image and the center line midpoint (Cp2). The center line midpoint (Cp2) may be a point that bisects the image portion of the center line (C0) captured in the identification image in the longitudinal direction. If the center line midpoint (Cp2) deviates to the left or right relative to the center point (Cp1), the control calculation unit (140) can generate a left-right correction value (Lc) that includes the direction of deviation and the degree of deviation.

[0083] Additionally, the control calculation unit (140) can generate a direction correction value (Dc) based on the difference in the angle of intersection or inclination between the center line (C0) and the center vertical line (C1) captured in the identification image. The center vertical line (C1) may be a line segment connecting the midpoint of the upper side and the midpoint of the lower side of the identification image. If the image portion of the center line (C0) is tilted with respect to the center vertical line (C1), this may mean that the direction of travel of the autonomous flight drone (100) is not parallel to the longitudinal direction of the center line (C0). The control calculation unit (140) can generate a direction correction value (Dc) based on this difference in inclination and transmit it to the flight control unit (110).

[0084] The flight control unit (110) can control the flight of the autonomous flight drone (100) using the altitude correction value (Hc), left / right correction value (Lc), and direction correction value (Dc) received from the control calculation unit (140). The flight control unit (110) can perform altitude control using the altitude correction value (Hc) so that the autonomous flight drone (100) flies forward while maintaining a distance (H) from the ceiling (11) of the underground power conduit (10). Additionally, the flight control unit (110) can perform left / right horizontal movement control using the left / right correction value (Lc) so that the autonomous flight drone (100) flies forward along the space below the centerline (C0). Additionally, the flight control unit (110) can perform horizontal angle (yaw) control using the direction correction value (Dc) so that the direction of travel of the autonomous flight drone (100) remains parallel to the longitudinal direction of the centerline (C0).

[0085] In the present invention, the autonomous flight drone (100) can take off from the docking station (300) and move to the inspection start point. The docking station (300) can be installed within the underground power conduit (10) and does not necessarily need to be located directly below the marker strip (200). Accordingly, the storage unit (150) may store path information for the autonomous flight drone (100) to move from the docking station (300) to the inspection start point. The path information for the start point may include sequential flight commands, such as ascending, moving forward, turning, adjusting altitude, or adjusting position, after taking off from the docking station (300).

[0086] The flight control unit (110) can take off the autonomous flight drone (100) from the docking station (300) according to an inspection command or a predetermined inspection schedule. After takeoff, the flight control unit (110) can control the autonomous flight drone (100) to move toward the inspection start point using the path information to reach the start point stored in the storage unit (150). During this process, the marker camera (130) takes an identification image toward the ceiling (11), and the control calculation unit (140) can search for a marker band (200), a center line (C0), or a start mark corresponding to the inspection start point in the identification image.

[0087] A start mark may be displayed at the inspection start point. Although not shown in the drawing, the start mark may consist of a shape, color, brightness, saturation, character, number, barcode, QR code, or geometric pattern that is distinguishable from a marker band (200), a center line (C0), or a position marker. When the control calculation unit (140) identifies the start mark in the identification image, it can determine that the autonomous flight drone (100) has reached the inspection start point.

[0088] When the control calculation unit (140) identifies the start mark, the flight control unit (110) can temporarily hover the autonomous flight drone (100) and perform alignment with respect to the marker band (200) and the centerline (C0). For example, the flight control unit (110) can adjust the altitude of the autonomous flight drone (100) so that it maintains a distance (H) from the ceiling (11) using an altitude correction value (Hc). Additionally, the flight control unit (110) can perform horizontal movement control in the left and right directions so that the autonomous flight drone (100) is positioned in the space below the centerline (C0) using a left-right correction value (Lc). Additionally, the flight control unit (110) can adjust the horizontal angle (yaw) so that the direction of travel of the autonomous flight drone (100) is parallel to the longitudinal direction of the centerline (C0) using a direction correction value (Dc). Once such alignment is complete, the flight control unit (110) can control the autonomous flight drone (100) to fly forward along the space below the marker strip (200) from the inspection start point to the inspection end point.

[0089] While the autonomous flight drone (100) is performing an inspection flight, the surveillance camera (120) can capture surveillance images of the underground power conduit (10) and the underground power distribution line (14). The captured surveillance images can be stored in the storage unit (150). At the same time, the marker camera (130) continues to capture identification images of the marker strip (200) and the centerline (C0), and the control calculation unit (140) can repeatedly generate altitude correction values ​​(Hc), left / right correction values ​​(Lc), and direction correction values ​​(Dc) using the identification images. The flight control unit (110) can use these correction values ​​to control the autonomous flight drone (100) to maintain a distance (H) from the ceiling (11), move along the space below the centerline (C0), and maintain a direction of travel parallel to the longitudinal direction of the centerline (C0).

[0090] A termination mark may be displayed at the inspection termination point. Although not shown in the drawing, the termination mark may be composed of a shape, color, brightness, saturation, character, number, barcode, QR code, or geometric pattern that is distinguishable from the start mark. When the control calculation unit (140) identifies the termination mark in the identification image, it can determine that the autonomous flight drone (100) has reached the inspection termination point.

[0091] When the control calculation unit (140) identifies the end mark, the flight control unit (110) can temporarily hover the autonomous flight drone (100) and control the autonomous flight drone (100) to change its direction of travel and return to the direction of the inspection start point. For example, the flight control unit (110) can rotate the direction of travel of the autonomous flight drone (100) by 180 degrees, then identify the marker band (200) and the center line (C0) again to align the left and right positions and horizontal angle (yaw) of the autonomous flight drone (100) to match the direction of return flight. Afterward, the autonomous flight drone (100) can return to the direction of the inspection start point along the space below the marker band (200).

[0092] When the autonomous flight drone (100) identifies the start mark again during the return flight, the flight control unit (110) can determine that the autonomous flight drone (100) has returned to the inspection start point. In this case, the autonomous flight drone (100) can be controlled to move from the inspection start point to the docking station (300) using the docking station return path information stored in the storage unit (150). The docking station return path information may include, for example, forward, turn, descend, position adjustment, and landing commands from the inspection start point to the docking station (300). Accordingly, the autonomous flight drone (100) can stably return to the docking station (300) even if the starting position of the marker band (200) and the position of the docking station (300) are separated from each other.

[0093] The docking station (300) may include a landing area where the autonomous flying drone (100) can land. When the autonomous flying drone (100) lands at the docking station (300), the docking station (300) can charge the battery of the autonomous flying drone (100). Charging can be performed through a contact charging terminal, a contactless charging device, or a combination thereof. Additionally, the communication unit (160) of the autonomous flying drone (100) can connect with the docking station (300) via a short-range communication network while the drone is landed at the docking station (300). At this time, the communication unit (160) can transmit surveillance video stored in the storage unit (150) to the docking station (300).

[0094] The docking station (300) can download surveillance footage from the autonomous flight drone (100). Additionally, the docking station (300) can download the flight log, path absence information, location identification information, distance traveled information, time of shooting information, or battery status information of the autonomous flight drone (100). The docking station (300) can store the downloaded data in its own storage unit or transmit it to a surveillance server (400) located at a remote location via a communication network (900). The communication network (900) may be a wired communication network, a wireless communication network, a mobile communication network, an optical communication network, or a combination thereof.

[0095] The monitoring server (400) receives monitoring images transmitted by the docking station (300) and analyzes the received monitoring images to diagnose whether there are abnormalities in the underground power conduit (10) and the underground power distribution line (14). For example, the monitoring server (400) can detect signs of overheating of the cable, insulation degradation, discoloration, damage, sagging, connection abnormalities, deformation or breakage of the cable hanger (13), leakage, flooding, condensation, ingress of foreign matter, corrosion, equipment damage, or fire based on the monitoring images or thermal images. The monitoring server (400) can determine whether there are abnormalities by using an image processing algorithm, an artificial intelligence analysis model, temperature distribution analysis, object detection, comparison of change amounts, or abnormal pattern analysis.

[0096] Additionally, the monitoring server (400) can identify the location where an abnormal sign was detected by using location identification information or distance traveled information received along with the monitoring video. Accordingly, the monitoring server (400) can indicate which section of the underground power conduit (10) the abnormality occurred in and can provide abnormality location information so that maintenance workers can quickly dispatch to the location.

[0097] In this way, according to the basic embodiment of the present invention, the autonomous flight drone (100) can take off from the docking station (300), move to the inspection starting point, and take surveillance images of the underground power conduit (10) and underground power distribution line (14) while controlling the altitude, left and right position and direction of travel based on the marker band (200) and centerline (C0). In addition, the autonomous flight drone (100) can return to the docking station (300) after the inspection is finished, transmit the surveillance images to the docking station (300), and charge the battery, and the docking station (300) can transmit the surveillance images to the monitoring server (400) to remotely diagnose whether there is an abnormality in the underground power conduit (10) and underground power distribution line (14).

[0098] Next, we examine an embodiment of the present invention in which a reference pixel (P0) is calculated and the altitude of an autonomous flight drone is controlled based on the difference between the reference pixel (P0) and the actual pixel (P1). FIG. 3 is a diagram explaining the principle of a marker camera capturing a marker band and a center line marked on the ceiling of an underground power conduit and calculating the number of pixels in the width direction of the marker band, and FIG. 4 is a diagram explaining the principle of controlling the altitude of an autonomous flight drone based on the difference between the reference pixel and the actual pixel.

[0099] Referring to FIG. 3, a marker camera (130) can photograph a marker band (200) and a center line (C0) marked on the ceiling (11) of an underground power conduit (10). Since the marker band (200) has a constant width (D) in actual space, when an autonomous flying drone (100) photographs the marker band (200) while positioned at a constant distance (H) from the ceiling (11), a number of pixels in the width direction corresponding to the width (D) of the marker band (200) appears in the identification image. In the present invention, the number of pixels in the width direction of the marker band (200) appearing in the identification image when the autonomous flying drone (100) is positioned at a target distance (H) can be defined as a reference pixel (P0).

[0100] Here, the separation distance (H) may be the distance from the ceiling surface of the underground power conduit (10) or the plane on which the marker strip (200) is marked to the optical center of the lens of the marker camera (130). However, according to the embodiment, the separation distance (H) may be defined as the distance from the center of the body of the autonomous flight drone (100), the installation reference point of the marker camera (130), or a reference point set on the drone to the ceiling surface. In this case, the reference pixel (P0) stored in the storage unit (150) may be calculated or corrected in advance according to the definition of the reference point.

[0101] In this embodiment, the specifications of the marker camera (130) may include a focal length (f), a horizontal length (W) of the image sensor, and a total number of horizontal pixels (Pw) of the image sensor. At this time, it is preferable that the focal length (f), the horizontal length (W) of the image sensor, the width (D) and the spacing (H) of the marker band (200) be converted and used in the same length unit. The total number of horizontal pixels (Pw) of the image sensor may correspond to the total number of horizontal pixels of the identification image.

[0102] When the marker camera (130) includes a lens of a general center projection method or provides an image corrected by a corresponding center projection model, the image width formed on the image sensor by the marker band (200) having an actual width (D) at the target separation distance (H) can be determined according to the ratio of the focal length (f) and the separation distance (H). Accordingly, the reference pixel (P0) can be calculated as follows using the width (D) of the marker band (200), the separation distance (H), the focal length (f), the horizontal length (W) of the image sensor, and the total number of horizontal pixels (Pw) of the image sensor.

[0103]

[0104] In the above equation, f·D / H may correspond to the horizontal length of the image formed on the image sensor by the width (D) of the marker band (200) at the target separation distance (H), and Pw / W may correspond to the number of pixels per unit length of the image sensor. Therefore, by multiplying these, the number of pixels corresponding to the width (D) of the marker band (200) at the target separation distance (H), i.e., the reference pixel (P0), can be obtained.

[0105] The reference pixel (P0) calculated in this manner may be stored in the storage unit (150). The reference pixel (P0) may be theoretically calculated based on the width (D) of the marker band (200), the target separation distance (H), and the specifications of the marker camera (130), or it may be stored as an experimentally corrected value by photographing the marker band (200) while the autonomous flying drone (100) is actually positioned at the target separation distance (H). Additionally, the reference pixel (P0) may be stored in a form that includes a correction value by considering the lens distortion of the marker camera (130), the installation angle, an image correction algorithm, or the installation state of the ceiling surface.

[0106] While the autonomous flying drone (100) is actually flying, the marker camera (130) repeatedly captures identification images of the marker band (200), and the control calculation unit (140) can calculate the number of pixels in the width direction of the marker band (200) captured in the identification images and calculate the actual pixels (P1). For example, the control calculation unit (140) can detect the first boundary line (B1) and the second boundary line (B2) of the marker band (200) in the identification images and calculate the number of pixels in the width direction between the first boundary line (B1) and the second boundary line (B2) as the actual pixels (P1).

[0107] Referring to FIG. 4, when the autonomous flight drone (100) is located at the target separation distance (H), the actual pixel (P1) of the marker band (200) captured in the identification image may appear to be the same as or nearly the same as the reference pixel (P0). In this case, the control calculation unit (140) can determine that the autonomous flight drone (100) is maintaining the target separation distance (H), and the flight control unit (110) can maintain the current altitude.

[0108] On the other hand, when the autonomous flying drone (100) approaches the ceiling (11), the distance between the marker camera (130) and the marker band (200) decreases, so the number of pixels in the width direction of the marker band (200) appearing in the identification image increases. That is, the actual pixel (P1) becomes larger than the reference pixel (P0). In this case, the control calculation unit (140) can determine that the autonomous flying drone (100) is excessively close to the ceiling (11).

[0109] Additionally, when the autonomous flying drone (100) moves away from the ceiling (11), the distance between the marker camera (130) and the marker band (200) increases, so the number of pixels in the width direction of the marker band (200) appearing in the identification image decreases. That is, the actual pixel (P1) becomes smaller than the reference pixel (P0). In this case, the control calculation unit (140) can determine that the autonomous flying drone (100) has deviated below the target separation distance (H).

[0110] In this embodiment, the altitude correction value (Hc) can be calculated as the value obtained by subtracting the actual pixel (P1) from the reference pixel (P0), i.e., P0-P1. Alternatively, the altitude correction value (Hc) may be calculated as the value obtained by multiplying P0-P1 by a predetermined proportionality coefficient, correction coefficient, or control gain.

[0111] For example, if the actual pixel (P1) is larger than the reference pixel (P0), P0-P1 becomes a negative value. In this case, since the altitude correction value (Hc) has a negative value, the flight control unit (110) can perform descent control so that the autonomous flight drone (100) moves away from the ceiling (11). Conversely, if the actual pixel (P1) is smaller than the reference pixel (P0), P0-P1 becomes a positive value. In this case, since the altitude correction value (Hc) has a positive value, the flight control unit (110) can perform ascent control so that the autonomous flight drone (100) moves closer to the ceiling (11).

[0112] The flight control unit (110) can maintain the current altitude of the autonomous flight drone (100) when the altitude correction value (Hc) is within a predetermined allowable range. Here, the allowable range may be set by considering the shooting error of the marker band (200), noise of the identification image, the resolution of the marker camera (130), fine vibrations occurring during flight, or control stability. For example, if the absolute value of the altitude correction value (Hc) is less than or equal to a predetermined reference value, it can be determined that the autonomous flight drone (100) is substantially maintaining the target separation distance (H).

[0113] If the altitude correction value (Hc) is a negative value outside the allowable range, the flight control unit (110) can control the propulsion unit or rotor output of the autonomous flight drone (100) to perform descent control. In this case, the autonomous flight drone (100) moves away from the ceiling (11), thereby reducing the risk of collision with the ceiling (11) or structures around the ceiling.

[0114] When the altitude correction value (Hc) is a positive value outside the allowable range, the flight control unit (110) can control the propulsion unit or rotor output of the autonomous flight drone (100) to perform ascent control. In this case, the autonomous flight drone (100) moves in a direction closer to the ceiling (11), so the target separation distance (H) can be restored. Accordingly, the autonomous flight drone (100) can stably photograph the marker strip (200) and the centerline (C0), and the shooting position of the surveillance video of the underground power distribution line (14) can also be maintained relatively consistently.

[0115] Additionally, the flight control unit (110) can increase the amount of control for ascent or descent control as the absolute value of the altitude correction value (Hc) increases. For example, if the absolute value of the altitude correction value (Hc) is large, it is determined that the autonomous flight drone (100) has significantly deviated from the target separation distance (H), and the ascent speed or descent speed can be increased; if the absolute value of the altitude correction value (Hc) is small, the ascent speed or descent speed can be reduced to perform fine correction. Accordingly, the autonomous flight drone (100) can quickly return to the target separation distance (H) while reducing shaking caused by rapid altitude changes.

[0116] According to the present embodiment, the autonomous flight drone (100) can determine the change in relative distance from the ceiling (11) by using the number of pixels in the width direction of the marker band (200) in the identification image captured by the marker camera (130), without relying solely on a separate precision distance sensor. In particular, since the actual width (D) of the marker band (200) is fixed, the change in the actual pixel (P1) can correspond to the change in distance between the autonomous flight drone (100) and the ceiling (11). Therefore, the altitude control of the autonomous flight drone (100) can be stably performed even in an environment where the illumination is low, such as inside an underground power conduit (10), and where there is dust, condensation, reflection, or interference from structures.

[0117] Next, we examine an embodiment in which a left-right correction value (Lc) is generated based on the relative position of a center point (Cp1) and a centerline midpoint (Cp2) in the present invention, and the horizontal movement of an autonomous flying drone in the left-right direction is controlled. FIG. 5 is a diagram illustrating the principle of calculating a left-right correction value using a center point, a centerline midpoint, a number of relative pixels in the left-right direction, and a left-right deviation distance in an identification image.

[0118] Referring to FIG. 5, a marker camera (130) can generate an identification image by photographing a marker strip (200) and a center line (C0) marked on the ceiling (11) of an underground power conduit (10). The identification image may include an image portion of the marker strip (200) and an image portion of the center line (C0), and a control calculation unit (140) can analyze the identification image to determine whether an autonomous flying drone (100) is located in the space below the center line (C0).

[0119] A center point (Cp1) may be defined at the exact center of the identification image. The center point (Cp1) may be the point where the horizontal center and the vertical center of the identification image meet, and may correspond to the current position of the autonomous flying drone (100) or the position where the optical axis of the marker camera (130) is projected onto the ceiling (11) when the lens optical axis of the marker camera (130) is maintained in a vertical direction toward the ceiling (11).

[0120] Additionally, the control calculation unit (140) can extract an image portion of the centerline (C0) captured in the identification image and calculate the point that bisects the image portion of the centerline (C0) in the length direction as the centerline midpoint (Cp2). For example, if the image portion of the centerline (C0) extends in the vertical direction of the identification image, the control calculation unit (140) can set the midpoint of the length segment detected within the identification image among the image portion of the centerline (C0) as the centerline midpoint (Cp2).

[0121] When the autonomous flying drone (100) is flying along the space directly below the centerline (C0), the center point (Cp1) of the identification image and the center point (Cp2) of the centerline may coincide with each other or appear at substantially the same location. In this case, the control calculation unit (140) can determine that the autonomous flying drone (100) has not deviated in the left or right direction, and the flight control unit (110) can maintain the current left or right position of the autonomous flying drone (100).

[0122] On the other hand, if the autonomous flying drone (100) deviates to the left or right from the space below the center line (C0), the center line midpoint (Cp2) in the identification image appears to deviate to the left or right relative to the center point (Cp1). In this case, the control calculation unit (140) can calculate the number of left and right relative pixels (P2) between the center point (Cp1) and the center line midpoint (Cp2).

[0123] The left-right relative pixel count (P2) may be the pixel distance between the center point (Cp1) and the centerline midpoint (Cp2) based on the horizontal direction of the identification image. According to an embodiment, the left-right relative pixel count (P2) is calculated as an absolute value, and deviation direction information indicating whether the centerline midpoint (Cp2) is located to the left or to the right of the center point (Cp1) may be separately assigned. Alternatively, the left-right relative pixel count (P2) may be calculated as a signed value with the left side as the negative direction and the right side as the positive direction based on the center point (Cp1).

[0124] The control calculation unit (140) can calculate the left-right deviation distance (d1) of the autonomous flying drone (100) using the width (D) of the marker band (200), the actual pixels (P1), and the number of left-right relative pixels (P2). Here, the actual pixels (P1) may be the number of pixels in the width direction of the marker band (200) captured in the identification image. Since the actual width (D) of the marker band (200) is predetermined, the actual pixels (P1) in the identification image become the number of pixels in the image corresponding to the actual width (D). Therefore, by converting the number of left-right relative pixels (P2) into an actual distance, the left-right deviation distance (d1) of the autonomous flying drone (100) can be obtained.

[0125] For example, the control calculation unit (140) can calculate the left and right deviation distance (d1) according to the following equation.

[0126]

[0127] In the above equation, D / P1 can be viewed as a conversion factor indicating how much length one pixel in the identification image corresponds to in actual space. Therefore, by multiplying the conversion factor by the number of left-right relative pixels (P2) between the center point (Cp1) and the center line center point (Cp2), a left-right deviation distance (d1) can be calculated, which indicates how much the autonomous flying drone (100) has deviated in the left-right direction from the space below the center line (C0).

[0128] The control calculation unit (140) can generate a left-right correction value (Lc) including a left-right deviation distance (d1) and a deviation direction, and transmit it to the flight control unit (110). For example, if the centerline center point (Cp2) is located to the right of the center point (Cp1), the control calculation unit (140) can determine that the autonomous flight drone (100) is in a state of deviating to the left of the centerline (C0), or at least that the centerline (C0) is located to the right of the autonomous flight drone (100). In this case, the left-right correction value (Lc) can be generated as a correction value for horizontally moving the autonomous flight drone (100) to the right.

[0129] Conversely, if the centerline center point (Cp2) is located to the left of the center point (Cp1), the control calculation unit (140) can determine that the centerline (C0) is located to the left of the autonomous flight drone (100). In this case, the left-right correction value (Lc) can be generated as a correction value to horizontally move the autonomous flight drone (100) to the left. However, if there is a difference between the left-right direction of the identification image and the actual left-right direction of the drone due to the camera mounting direction or image inversion processing, the correction direction can be determined using the camera installation direction information or image coordinate transformation information stored in the storage unit (150).

[0130] The flight control unit (110) can perform left-right horizontal movement control of the autonomous flight drone (100) using left-right correction values ​​(Lc). Specifically, the flight control unit (110) can move the autonomous flight drone (100) to the left or right so that the centerline center point (Cp2) in the identification image approaches the center point (Cp1). At this time, the flight control unit (110) can perform left-right horizontal movement using roll control, left-right thrust control, rotor output differential control, or attitude control of the autonomous flight drone (100).

[0131] Additionally, the flight control unit (110) can vary the amount of left-right movement control depending on the magnitude of the left-right deviation distance (d1). For example, when the left-right deviation distance (d1) is large, the autonomous flight drone (100) is significantly deviated from the space below the centerline (C0), so the flight control unit (110) can apply a relatively large amount of left-right movement control. Conversely, when the left-right deviation distance (d1) is small, a fine amount of left-right movement control can be applied to reduce shaking of the autonomous flight drone (100) and align it with the centerline (C0).

[0132] Additionally, the flight control unit (110) can maintain the current left and right position of the autonomous flight drone (100) when the left and right correction value (Lc) is within a predetermined allowable range. The allowable range can be set by considering the resolution of the marker camera (130), noise in the identification image, recognition error of the marker band (200) or centerline (C0), and the fine vibration and control stability of the autonomous flight drone (100). Accordingly, even if the centerline center point (Cp2) and the center point (Cp1) do not completely coincide, unnecessary left and right correction can be suppressed within a range where it can be seen as flying along the space below the centerline (C0).

[0133] In this embodiment, the number of relative pixels in the left and right directions (P2) is converted into an actual deviation distance (d1) using the actual pixels (P1), so even if the altitude of the autonomous flying drone (100) changes temporarily, the deviation distance (d1) in the left and right directions can be calculated by reflecting the image scale at that point in time. For example, when the autonomous flying drone (100) approaches the ceiling (11) and the number of pixels in the width direction of the marker band (200) increases, the actual pixels (P1) increase, so the actual deviation distance (d1) corresponding to the same relative pixels in the left and right directions (P2) is calculated to be relatively small. Conversely, when the autonomous flying drone (100) moves away from the ceiling (11) and the actual pixels (P1) decrease, the actual deviation distance (d1) corresponding to the same relative pixels in the left and right directions (P2) is calculated to be relatively large. Therefore, compared to a method that performs left and right correction based only on the difference in pixels in the image, this embodiment can more accurately reflect the degree of left and right deviation in actual space.

[0134] In this way, according to the present invention, the autonomous flight drone (100) can determine whether there is a deviation in the left and right directions from the space below the center line (C0) by using the relative position of the center point (Cp1) of the identification image and the center point (Cp2) of the center line. In addition, since the left and right deviation distance (d1) can be calculated using the actual width (D) of the marker strip (200) and the actual pixel (P1), the autonomous flight drone (100) can fly stably along the space below the center line (C0) even in an environment that is narrow, such as inside an underground power conduit (10), and where cable hangers (13), underground power distribution lines (14), and wall structures exist.

[0135] Next, we examine an embodiment in which a direction correction value (Dc) is generated based on the difference in the intersection angle or slope between the image portion of the center line (C0) and the center vertical line (C1) in the present invention, and the horizontal angle (yaw) of an autonomous flying drone is controlled. FIG. 5 is a diagram illustrating the principle of calculating a direction correction value using the center line (C0), the center vertical line (C1), the intersection angle (θ), and the relative coordinates (x, y) in an identification image.

[0136] Referring to FIG. 5, a marker camera (130) can generate an identification image by capturing a marker strip (200) and a center line (C0) marked on the ceiling (11) of an underground power conduit (10). Since the center line (C0) is located at the center of the width direction of the marker strip (200) and extends along the length direction of the marker strip (200), when an autonomous flying drone (100) is flying in a direction parallel to the length direction of the center line (C0), the image portion of the center line (C0) captured in the identification image may appear parallel to or substantially coincide with the center vertical line (C1) of the identification image.

[0137] The center vertical line (C1) may be a line segment connecting the midpoint of the upper side and the midpoint of the lower side of the identification image. The center vertical line (C1) may be used as a reference line indicating the direction of travel of the autonomous flight drone (100) in the image captured by the marker camera (130). Therefore, if the image portion of the center line (C0) appears tilted with respect to the center vertical line (C1), this may mean that the direction of travel of the autonomous flight drone (100) is not parallel to the length direction of the center line (C0) and is tilted in one direction.

[0138] The control calculation unit (140) can detect the image portion of the center line (C0) in the identification image and calculate the intersection angle (θ) formed by the image portion of the center line (C0) and the center vertical line (C1). Here, the intersection angle (θ) may be a value indicating how much the image portion of the center line (C0) is tilted clockwise or counterclockwise with respect to the center vertical line (C1). According to an embodiment, the control calculation unit (140) may calculate the difference in inclination between the image portion of the center line (C0) and the center vertical line (C1) and convert the difference in inclination into the intersection angle (θ).

[0139] As illustrated in FIG. 5(c), the control calculation unit (140) can set the intersection point of the image portion of the center line (C0) and the center vertical line (C1) as the coordinate origin (0,0). Then, an arbitrary point other than the intersection point can be selected among the image portions of the center line (C0), and the relative coordinates for the arbitrary point can be defined as x and y. In this case, the control calculation unit (140) can calculate the intersection angle (θ) according to the following equation.

[0140]

[0141] Here, x can represent the degree to which an arbitrary point selected from the image portion of the center line (C0) is offset in the left-right direction with respect to the coordinate origin (0,0), and y can represent the degree to which the arbitrary point is offset in the vertical direction with respect to the coordinate origin (0,0). Therefore, x / y can correspond to the slope that the image portion of the center line (C0) has with respect to the center vertical line (C1), and tan -1 (x / y) can correspond to the intersection angle (θ) between the image portion of the center line (C0) and the center vertical line (C1).

[0142] According to the embodiment, the image portion of the center line (C0) and the center vertical line (C1) may not directly intersect within the identification image. For example, if the error in the direction of travel of the autonomous flight drone (100) is large, or if the image portion of the center line (C0) is captured off-center to one side of the identification image, the two line segments may not meet each other within the identification image. Even in this case, the control calculation unit (140) can set a virtual intersection point where the image portion of the center line (C0) and the extension of the center vertical line (C1) meet as the coordinate origin (0,0), and calculate the intersection angle (θ) based on the virtual intersection point.

[0143] The control calculation unit (140) can generate a calculated intersection angle (θ) or a value proportional to the intersection angle (θ) as a direction correction value (Dc). For example, the direction correction value (Dc) may be the intersection angle (θ) itself, or it may be a value obtained by multiplying the intersection angle (θ) by a predetermined proportionality coefficient, control gain, or correction coefficient. Additionally, the direction correction value (Dc) may be a control value indicating how much and in which direction the horizontal angle (yaw) of the autonomous flight drone (100) will be corrected.

[0144] In this embodiment, if the image portion of the center line (C0) is tilted clockwise with respect to the center vertical line (C1), the direction correction value (Dc) may have a positive value, and if the image portion of the center line (C0) is tilted counterclockwise with respect to the center vertical line (C1), the direction correction value (Dc) may have a negative value. However, in actual control, the meaning of the positive and negative values ​​may be set in reverse depending on the mounting direction of the marker camera (130), the setting of the image coordinate system, whether left-right inversion processing is performed, or the definition of the yaw control direction of the flight control unit (110). In this case, the storage unit (150) may store conversion information between the image coordinate system and the drone control coordinate system, and the control calculation unit (140) may determine the sign of the direction correction value (Dc) using the conversion information.

[0145] The flight control unit (110) receives a direction correction value (Dc) from the control calculation unit (140) and can control the horizontal angle (yaw) of the autonomous flight drone (100) using the direction correction value (Dc). Specifically, when the direction correction value (Dc) is within a predetermined allowable range, the flight control unit (110) can maintain the current angle of travel of the autonomous flight drone (100). Here, the allowable range can be set by considering the resolution of the marker camera (130), the detection error of the centerline (C0), the noise of the identification image, the minute vibration of the autonomous flight drone (100), and the yaw control stability.

[0146] On the other hand, if the direction correction value (Dc) is outside the allowable range, the flight control unit (110) can perform horizontal angle (yaw) control for the autonomous flight drone (100) so that the intersection angle (θ) approaches 0. That is, the flight control unit (110) can correct the heading of the autonomous flight drone (100) in a direction such that the image portion of the center line (C0) in the identification image is parallel to or coincides with the center vertical line (C1). Accordingly, the direction of travel of the autonomous flight drone (100) can be restored to a state parallel to the longitudinal direction of the center line (C0).

[0147] For example, if the image portion of the center line (C0) in the identification image is tilted clockwise with respect to the center vertical line (C1), the control calculation unit (140) can generate a positive direction correction value (Dc). At this time, the flight control unit (110) can correct the horizontal angle (yaw) in the direction corresponding to the positive direction correction value (Dc) to control the image portion of the center line (C0) to be closer to the center vertical line (C1). Conversely, if the image portion of the center line (C0) is tilted counterclockwise, the control calculation unit (140) can generate a negative direction correction value (Dc), and the flight control unit (110) can correct the horizontal angle (yaw) in the direction corresponding to it.

[0148] Additionally, the flight control unit (110) can increase the amount of control for horizontal angle (yaw) control as the absolute value of the direction correction value (Dc) increases. For example, when the intersection angle (θ) is large, the direction of travel of the autonomous flight drone (100) is significantly deviated from the longitudinal direction of the centerline (C0), so the flight control unit (110) can apply a relatively large amount of yaw control to quickly correct the direction of travel. Conversely, when the intersection angle (θ) is small, a small amount of yaw control can be applied to perform fine correction while preventing the autonomous flight drone (100) from rotating or vibrating excessively.

[0149] Horizontal angle (yaw) control based on the direction correction value (Dc) can be performed together with horizontal movement control based on the left-right correction value (Lc). For example, when the centerline center point (Cp2) is deviated to the left or right from the center point (Cp1) and the image portion of the centerline (C0) is tilted relative to the center vertical line (C1), the flight control unit (110) can correct the left-right position of the autonomous flight drone (100) using the left-right correction value (Lc) and correct the horizontal angle (yaw) of the autonomous flight drone (100) using the direction correction value (Dc). Accordingly, the autonomous flight drone (100) can be positioned in the space below the centerline (C0) while simultaneously maintaining a direction of travel parallel to the longitudinal direction of the centerline (C0).

[0150] According to the present embodiment, the autonomous flying drone (100) does not merely identify the marker band (200) and position itself below it, but can determine how far its direction of travel deviates from the longitudinal direction of the center line (C0) by using the difference in the angle of intersection or slope between the image portion of the center line (C0) and the center vertical line (C1). Therefore, it can suppress the accumulation of yaw errors while performing long-distance flight inside the underground power conduit (10), and can perform stable path-following flight while reducing the risk of collision with cable hangers (13), underground power distribution lines (14), walls, or other structures.

[0151] Next, we examine an embodiment of the present invention that controls the entry and passage of an autonomous drone into and through a curved section using a curve marker marked on a curved section of an underground power conduit. FIG. 6 is a diagram illustrating the principle of controlling the flight of an autonomous drone in a curved section using a marker band, a centerline, and a curve marker marked on a curved section according to an embodiment of the present invention.

[0152] Referring to FIG. 6, the underground power conduit (10) may include not only straight sections but also curved sections. In the curved sections of the underground power conduit (10), the marker strip (200) marked on the ceiling (11) may also be extended in a curved shape corresponding to the curved shape of the underground power conduit (10), and the center line (C0) marked at the center of the width direction of the marker strip (200) may also be extended in a curved shape.

[0153] In a section where the center line (C0) extends in a curved shape as described above, it may be difficult to pass through the curved section stably by controlling the entire image portion of the center line (C0) to be aligned with or parallel to the center vertical line (C1) of the identification image, as is the case in a straight section. Therefore, in this embodiment, a curve marker (210) is displayed at the starting position and the ending position of the curved section, respectively, and the autonomous flying drone (100) can determine the entry or exit of the curved section by identifying the curve marker (210).

[0154] The curved marker (210) may be placed within the area of ​​the marker band (200). For example, the curved marker (210) may be displayed in the inner area in the width direction of the marker band (200), on one side of the center line (C0), or in an area that does not overlap with the center line (C0). It is preferable that the curved marker (210) be formed in a shape that can be distinguished from the marker band (200) and the center line (C0) in the identification image captured by the marker camera (130) of the autonomous flying drone (100).

[0155] The curve marker (210) may be formed from a QR code, a barcode, a color pattern, a brightness pattern, a shape pattern, a character, a number, or a combination thereof. Additionally, the curve marker (210) may include information regarding an identification ID, the radius of curvature (R) of the curved section, and the direction of curvature when entering the curved section based on the curve marker (210). Here, the direction of curvature may be information indicating whether the centerline (C0) curves to the left or the right when the autonomous flying drone (100) passes the curve marker (210) and enters the curved section.

[0156] The radius of curvature (R) may refer to the radius of curvature of the reference curve of the centerline (C0) or marker band (200) in the curved section. When the radius of curvature (R) is large, the curved section can be seen as being gently curved, and when the radius of curvature (R) is small, the curved section can be seen as being relatively steeply curved. Therefore, the autonomous flying drone (100) can use the radius of curvature (R) to adjust the amount of correction for the direction of travel, the amount of correction for the horizontal angle (yaw), or the forward speed in the curved section.

[0157] The marker camera (130) can capture identification images of the marker strip (200), centerline (C0), and curve marker (210) while the autonomous flying drone (100) is flying. The control calculation unit (140) receives the identification images from the marker camera (130) and can determine whether the curve marker (210) is identified in the identification images. The control calculation unit (140) can extract information contained in the curve marker (210) by analyzing the shape, color, pattern, character, number, or code information of the curve marker (210) within the identification images.

[0158] When a curve marker (210) is identified in an identification image, the control calculation unit (140) can generate curve section entry information or curve section end information using the identification ID, radius of curvature (R), and curvature direction included in the curve marker (210). The curve section entry information may be information indicating that the autonomous flight drone (100) must enter a curve section, and the curve section end information may be information indicating that the autonomous flight drone (100) must exit the curve section and switch to a straight section or a normal path following state.

[0159] The curve section entry information may include the identification ID of the curve marker (210), the radius of curvature (R), the direction of curvature, the length of the curve section, the recommended travel speed, the recommended yaw control gain, or the curve section passing mode switching information. However, the length of the curve section, the recommended travel speed, or the recommended yaw control gain does not necessarily have to be directly included in the curve marker (210), and may be provided in a manner in which the control calculation unit (140) retrieves the curve section information and the identification ID that are pre-stored in the storage unit (150).

[0160] When the flight control unit (110) receives information on entering a curved section from the control calculation unit (140), it can switch the flight mode of the autonomous flight drone (100) to a curved section passing mode. In the curved section passing mode, the flight control unit (110) can perform direction correction based on the portion of the image portion of the centerline (C0) located in front of the autonomous flight drone (100) in the direction of travel.

[0161] Specifically, since the entire image portion of the centerline (C0) can be captured in a curved shape in the curved section, the control calculation unit (140) can calculate the tangent direction or local length direction of a portion of the image portion of the centerline (C0) located in the forward direction of travel, rather than calculating the intersection angle with the center vertical line (C1) by viewing the entire image portion of the centerline (C0) as a single straight line. Here, the portion located in the forward direction of travel may correspond to the portion of the centerline (C0) that the autonomous flight drone (100) must follow in the identification image at the next moment.

[0162] For example, when the autonomous flight drone (100) is flying forward from the inspection start point to the inspection end point, a local section of the centerline (C0) can be selected in the area corresponding to the forward direction of travel in the identification image. The control calculation unit (140) can calculate the tangential direction or local length direction of the selected local section and generate a direction correction value (Dc) so that the tangential direction or local length direction is maintained parallel to the center vertical line (C1) of the identification image.

[0163] The flight control unit (110) can control the horizontal angle (yaw) of the autonomous flight drone (100) based on this direction correction value (Dc). Accordingly, the autonomous flight drone (100) can correct the horizontal angle by reflecting not only the current position of the centerline (C0) in the curved section but also the direction in which the centerline (C0) bends in the forward direction of travel. As a result, the autonomous flight drone (100) can fly while naturally turning along the longitudinal direction of the centerline (C0) even in the curved section.

[0164] Additionally, the flight control unit (110) can correct the direction of travel of the autonomous flight drone (100) to correspond to the radius of curvature (R) and the direction of curvature included in the curve marker (210). For example, if the direction of curvature is to the left and the radius of curvature (R) is set small, the flight control unit (110) can correct the horizontal angle (yaw) of the autonomous flight drone (100) to the left to a relatively large degree, or reduce the forward speed to pass through the curved section stably. Conversely, if the direction of curvature is to the right and the radius of curvature (R) is set large, the flight control unit (110) can control the autonomous flight drone (100) to fly along the centerline (C0) while performing gentle yaw control in the right direction.

[0165] The flight control unit (110) can use the altitude correction value (Hc) and the left / right correction value (Lc) together even after receiving information on entering the curved section. That is, the autonomous flight drone (100) can maintain the distance (H) from the ceiling (11) based on the number of pixels in the width direction of the marker strip (200) while passing through the curved section, and can perform left / right horizontal movement control to fly along the space below the center line (C0) based on the relative position of the center point (Cp1) and the center line center point (Cp2). Accordingly, the altitude and left / right position can be stably maintained together even while horizontal angle (yaw) control is performed in the curved section.

[0166] The control calculation unit (140) can generate curve section end information when it identifies a curve marker (210) marked at the end point of the curve section. The curve section end information may be information that causes the autonomous flight drone (100) to stop the correction of the direction of travel corresponding to the curve section and return to normal forward flight control along the length direction of the centerline (C0).

[0167] When the flight control unit (110) receives curve section end information from the control calculation unit (140), it can terminate the correction of the direction of travel based on the radius of curvature (R) and the bending direction that was performed in the curve section passing mode. Subsequently, the flight control unit (110) can generate a direction correction value (Dc) based on the difference in the intersection angle or slope between the image portion of the center line (C0) captured in the identification image and the center vertical line (C1), and control the autonomous flight drone (100) to fly forward along the longitudinal direction of the center line (C0).

[0168] At this time, immediately after the end of the curve section, the direction of travel of the autonomous flight drone (100) may not completely coincide with the direction of the straight section of the centerline (C0). Therefore, the flight control unit (110) can gradually change the control gain of the direction correction value (Dc) for a certain period of time or a certain distance after receiving the information on the end of the curve section. For example, instead of immediately switching from the curve section correction mode to the general straight section following mode, the amount of yaw control can be gradually reduced and continuously corrected so that the image portion of the centerline (C0) approaches the center vertical line (C1). Accordingly, the sudden attitude change or vibration of the autonomous flight drone (100) at the end of the curve section can be reduced.

[0169] Additionally, even when the autonomous flight drone (100) performs a return flight by changing its direction of travel by 180 degrees at the inspection end point, the flight control unit (110) can interpret the information of the curve marker (210) by considering whether the current flight state is an inspection flight state or a return flight state. For example, the curve marker (210) that was interpreted as the starting position of the curve section in the inspection flight state can correspond to the ending position of the curve section in the return flight state. Accordingly, the identification ID of the curve marker (210) and the curve section information can be stored in the storage unit (150) by matching them according to the flight direction, and the control calculation unit (140) can generate curve section entry information or curve section end information by considering the current flight state and the identification ID together.

[0170] According to the present embodiment, the autonomous flight drone (100) can determine the direction of turning in the curved section of the underground power conduit (10) by using curve markers (210) marked at the starting and ending positions of the curved section, rather than simply using the shape of the obstacle or wall in front. In addition, since the horizontal angle (yaw) of the autonomous flight drone (100) can be corrected using the radius of curvature (R) and the bending direction included in the curve markers (210), stable autonomous flight can be performed based on the marker strip (200) and the centerline (C0) even if the underground power conduit (10) includes a curved section.

[0171] Next, we examine an embodiment in which the start and end of a curved section are determined using the identification sequence of curve markers in the present invention. FIG. 6 is a diagram illustrating the principle of controlling the flight of an autonomous drone in a curved section using a marker band, a centerline, and curve markers marked on a curved section according to an embodiment of the present invention.

[0172] Referring to FIG. 6, a marker strip (200) may be marked on the ceiling (11) of the underground power conduit (10), and a center line (C0) may be marked at the center of the width direction of the marker strip (200). The marker strip (200) and the center line (C0) may be extended in a straight line shape in the straight section of the underground power conduit (10), and may be extended in a curved shape corresponding to the curved shape of the underground power conduit (10) in the curved section of the underground power conduit (10).

[0173] In this embodiment, a curve marker (210) may be displayed at each of the starting point and ending point positions of the curve section. The curve marker (210) may be placed within the area of ​​the marker band (200) and may be formed with a shape, color, brightness, saturation, geometric pattern, character, number, barcode, QR code, or a combination thereof that can be distinguished from the marker band (200) and the center line (C0).

[0174] The curve marker (210) may include information regarding an identification ID, a radius of curvature (R) of the curved section, and the direction of curvature when entering the curved section based on the curve marker (210). Here, the identification ID may be information for identifying the curve marker (210) or the curved section, the radius of curvature (R) may be information indicating the degree to which the centerline (C0) or marker band (200) bends in the curved section, and the direction of curvature may be information indicating whether the centerline (C0) bends to the left or the right when the autonomous flying drone (100) passes the curve marker (210) and enters the curved section.

[0175] While the autonomous flying drone (100) flies from the inspection start point to the inspection end point, the marker camera (130) can repeatedly capture identification images of the marker band (200), centerline (C0), and curve marker (210). The control calculation unit (140) receives the identification images from the marker camera (130) and can determine whether the curve marker (210) is identified in the identification images.

[0176] The control calculation unit (140) can count the identification order of curve markers (210) when curve markers (210) are sequentially identified in the identification image while the autonomous flight drone (100) is flying from the inspection start point to the inspection end point. For example, when the autonomous flight drone (100) starts an inspection flight from the inspection start point, the control calculation unit (140) can initialize the curve marker count value. Subsequently, whenever a new curve marker (210) is identified in the identification image, the control calculation unit (140) can increase the curve marker count value by 1.

[0177] At this time, it is preferable for the control calculation unit (140) not to count the same curve marker (210) redundantly even if the same curve marker (210) is continuously captured over multiple frames. For example, the control calculation unit (140) can prevent redundant counting of the same curve marker (210) by using the identification ID of the curve marker (210), the change in position within the identification image, the identification duration, or a list of previously counted marker IDs. Alternatively, the control calculation unit (140) may be configured to count the curve marker (210) once only when the curve marker (210) passes through a predetermined reference area of ​​the identification image.

[0178] The control calculation unit (140) can determine whether the curve marker (210) is a curve section start marker displayed at the starting position of the curve section or a curve section end marker displayed at the ending position of the curve section, depending on whether the identification order of the curve marker (210) is odd or even.

[0179] For example, when the autonomous flying drone (100) flies from the inspection start point to the inspection end point, a curve marker (210) at the starting point position and a curve marker (210) at the ending point position may be sequentially placed in one curve section. Accordingly, the control calculation unit (140) can determine the first identified curve marker (210) as the start marker of the first curve section and the second identified curve marker (210) as the end marker of the first curve section. Additionally, the third identified curve marker (210) can be determined as the start marker of the second curve section and the fourth identified curve marker (210) as the end marker of the second curve section.

[0180] That is, when a curve marker (210) with an odd identification order is identified, the control calculation unit (140) can determine that the curve marker (210) is a curve section start marker displayed at the starting position of the curve section. In this case, the control calculation unit (140) can extract the identification ID, radius of curvature (R), and bending direction included in the curve marker (210), and generate curve section entry information including the above.

[0181] The curve section entry information may be information indicating that the autonomous flight drone (100) has entered or is about to enter a curve section. The curve section entry information may include an identification ID of a curve marker (210), a radius of curvature (R), a bending direction, a time of entry into the curve section, a position of entry into the curve section, a command to switch to a curve section passing mode, or yaw control conditions to be applied in the curve section. When the flight control unit (110) receives the curve section entry information, it may perform horizontal angle (yaw) control of the autonomous flight drone (100) to correspond to the radius of curvature (R) and the bending direction.

[0182] Conversely, if a curve marker (210) with an even identification order is identified, the control calculation unit (140) may determine that the curve marker (210) is a curve section end marker displayed at the end point of the curve section. In this case, the control calculation unit (140) may generate curve section end information to terminate the direction of travel correction corresponding to the curve section.

[0183] The curve section end information may be information indicating that the autonomous flight drone (100) has passed through the curve section or reached the end position of the curve section. The curve section end information may include an identification ID of the ended curve section, a time of end of the curve section, a position of end of the curve section, a command to release the curve section passing mode, or a command to return to the general path following mode. When the flight control unit (110) receives the curve section end information, it may terminate the direction correction performed according to the radius of curvature (R) and the bending direction, and return to general direction control based on the length direction of the centerline (C0).

[0184] According to this configuration, the curve marker displayed at the starting position of the curve section and the curve marker displayed at the ending position do not necessarily have to have different appearances. The control calculation unit (140) can distinguish the odd-numbered curve markers as curve section start markers and the even-numbered curve markers as curve section end markers by utilizing the order in which the curve markers (210) are identified while the autonomous flying drone (100) flies from the inspection start point to the inspection end point. Accordingly, the autonomous flying drone (100) can stably determine the entry and exit of the curve section while simplifying the display structure of the curve markers (210).

[0185] Additionally, the control calculation unit (140) can supplement the accuracy of determining the identification order by using the identification ID of the curve marker (210) together. For example, the storage unit (150) may store arrangement information of curve sections from the inspection start point to the inspection end point, and the arrangement information of the curve sections may include the identification ID, placement order, radius of curvature (R), bending direction, or length of the curve section of the curve marker (210) corresponding to each curve section. The control calculation unit (140) can determine whether there is an identification error or a missing marker by comparing the identification order of the actually identified curve marker (210) with the arrangement information of the curve sections stored in the storage unit (150).

[0186] For example, the control calculation unit (140) may determine that a curve marker identification error or a path following error has occurred if an identification ID corresponding to an even-numbered curve marker is identified first while an odd-numbered curve section start marker that is to be identified is not identified. In this case, the control calculation unit (140) may request the flight control unit (110) to control speed reduction, hovering, re-identification, path realignment, or safe return.

[0187] Additionally, the control calculation unit (140) may determine that the passage through the curved section was not successfully completed if the curved section end marker is not identified within a certain time or distance after the curved section start marker has been identified. In this case, the control calculation unit (140) may transmit warning information to the flight control unit (110) or control the autonomous flight drone (100) to realign to the centerline (C0). Such auxiliary control may be applied as an optional embodiment to enhance safety in the curved section.

[0188] This embodiment may be described based on an inspection flight state in which an autonomous flight drone (100) flies from an inspection start point to an inspection end point. However, if the autonomous flight drone (100) changes its direction of travel at the inspection end point and performs a return flight toward the inspection start point, the meaning of the curve marker (210) can be interpreted based on the opposite direction using curve section arrangement information and current flight state information stored in the storage unit (150). For example, a curve marker (210) that was determined to be a curve section start marker in the inspection flight state can be interpreted as a marker for the end of the corresponding curve section in the return flight state. By considering flight direction information together in this way, the passage of the curve section can be controlled in both the inspection flight and the return flight directions using the same curve marker (210).

[0189] In this way, according to the present invention, the start and end of a curved section can be determined by using the identification sequence of the curve markers (210), so the autonomous flight drone (100) can distinguish the entry and end of the curved section without having to separately install a start-only marker and an end-only marker of a complex structure for each curved section. In addition, since the identification ID, radius of curvature (R), and bending direction included in the curve markers (210) can be used together, stable autonomous flight can be performed based on the marker strip (200) and centerline (C0) even in the curved section of the underground power conduit (10).

[0190] Next, we examine an embodiment in which, when the marker camera in the present invention includes a fisheye lens with an equidistant projection method, a reference pixel (P0) is calculated by reflecting the projection characteristics of the fisheye lens, and the altitude of an autonomous flying drone is controlled using this. FIG. 7 is a diagram illustrating the principle of calculating the number of pixels in the width direction of a marker band and performing altitude control using a marker camera including a fisheye lens with an equidistant projection method according to an embodiment of the present invention.

[0191] Referring to FIG. 7, the marker camera (130) may include an equidistant projection fisheye lens. An equidistant projection fisheye lens may be a lens in which the image position on the image sensor is determined in proportion to the angle of incidence of the incident light rays relative to the optical axis. That is, while the image size is determined according to the ratio between the actual width of the subject and the shooting distance in a standard center projection lens, in an equidistant projection fisheye lens, the pixel spacing in the image may be determined according to the angle formed by the two boundaries of the subject relative to the lens optical axis.

[0192] These fisheye lenses have a wider angle of view compared to standard lenses ( Since it can provide ), even when the autonomous flying drone (100) flies relatively close to the ceiling (11) of the underground power conduit (10), it can photograph the marker strip (200), centerline (C0), and surrounding identification targets over a wide range. Therefore, in environments where the flight space is narrow, such as inside the underground power conduit (10), and the distance between the marker camera (130) and the ceiling (11) is short, it may be preferable to use a fisheye lens.

[0193] In particular, the configuration using a fisheye lens of an equidistant projection method in this embodiment is advantageous when the ceiling height of the underground power conduit (10) is low, so the autonomous flying drone (100) must fly close to the ceiling (11) and precise altitude control is required. When the distance (H) between the ceiling (11) and the marker camera (130) is short, the two boundaries of the marker strip (200) occupy a relatively large angle range within the field of view of the marker camera (130). In this state, even if the altitude of the autonomous flying drone (100) changes only slightly, the field of view in the width direction of the marker strip (200) changes relatively significantly, and as a result, the actual measured pixel (P1) in the width direction of the marker strip (200) captured in the identification image can also change sensitively.

[0194] However, when using a fisheye lens, an error may occur if the reference pixel (P0) is calculated in the same way as a standard lens. In a standard center projection lens, the image width can be calculated based on the ratio of the width (D), separation distance (H), and focal length (f) of the marker band (200), but in an equidistant projection fisheye lens, it is preferable to first calculate the angle formed by the two boundaries of the marker band (200) with respect to the lens optical axis, and then calculate the image width on the image sensor by multiplying that angle by the focal length (f).

[0195] In this embodiment, the specifications of the marker camera (130) may include the focal length (f) of the fisheye lens, the horizontal length (W) of the image sensor, and the total number of horizontal pixels (Pw) of the image sensor. Additionally, the marker band (200) has a constant width (D), and a reference pixel (P0) can be calculated based on the case where the lens optical axis of the marker camera (130) is positioned vertically toward the center or centerline (C0) in the width direction of the marker band (200) while the autonomous flying drone (100) is located at a target separation distance (H).

[0196] At this time, since the first boundary line (B1) and the second boundary line (B2) are each spaced apart by D / 2 with respect to the center in the width direction of the marker band (200), the angle formed by one side boundary of the marker band (200) with respect to the lens optical axis is tan -1 It can be represented as {D / (2H)}. In the equidistant projection method, since the position on the image sensor is proportional to the angle of incidence, the length on the image sensor corresponding to the boundary between both sides of the marker band (200) is 2f·tan -1 It can be calculated as {D / (2H)}.

[0197] Accordingly, when the total number of pixels corresponding to the horizontal length (W) of the image sensor is Pw, the reference pixel (P0) can be calculated according to the following equation.

[0198]

[0199] In the above equation, tan -1 {D / (2H)} corresponds to the angle formed by one boundary with respect to the lens optical axis with respect to the center in the width direction of the marker band (200), and 2f·tan -1 {D / (2H)} can correspond to the image width of the marker band (200) formed on the image sensor by an equal-distance projection fisheye lens. Additionally, since Pw / W corresponds to the number of pixels per unit length of the image sensor, P0 calculated according to the above formula becomes the number of reference pixels corresponding to the width (D) of the marker band (200) at the target separation distance (H).

[0200] The reference pixel (P0) calculated in this manner can be stored in the storage unit (150). The reference pixel (P0) can be theoretically calculated using the focal length (f) of the fisheye lens, the horizontal length (W) of the image sensor, the total number of horizontal pixels (Pw) of the image sensor, the width (D) of the marker band (200), and the target separation distance (H). Additionally, according to the embodiment, the reference pixel (P0) may be stored as an experimentally corrected value by photographing the marker band (200) with the autonomous flying drone (100) positioned at the target separation distance (H) in an actual underground power conduit (10) or test environment.

[0201] When the marker camera (130) includes a fisheye lens of an equidistant projection method, the image portion of the marker band (200) captured in the identification image may appear as a curve or convex shape rather than a straight line shape due to lens distortion. In particular, as one moves further away from the center of the identification image, the boundary lines (B1, B2) on both sides of the marker band (200) may become curved, or even if the marker band (200) has the same actual width (D), the number of pixels in the width direction may appear different depending on the position within the image.

[0202] Accordingly, when the control calculation unit (140) calculates the actual pixel (P1), it can preferably calculate the number of pixels in the width direction of the marker band (200) on a predetermined horizontal reference line. For example, a center horizontal line (C3) may be defined in the identification image. The center horizontal line (C3) may be a horizontal reference line connecting the center point of the left side and the center point of the right side of the identification image. The control calculation unit (140) can detect the first boundary line (B1) and the second boundary line (B2) of the marker band (200) on the center horizontal line (C3), and calculate the number of pixels between the first boundary line (B1) and the second boundary line (B2) as the actual pixel (P1).

[0203] Alternatively, the control calculation unit (140) may calculate the number of pixels in the width direction of the marker band (200) on a horizontal reference line passing through the center point (Cp1) of the identification image. In this case, the center horizontal line (C3) may be defined as a horizontal reference line passing through the center point (Cp1). Additionally, according to an embodiment, the control calculation unit (140) may calculate the number of pixels in the width direction of the marker band (200) on a horizontal reference line passing through the center point (Cp2). By predetermining the calculation reference line for the actual pixel (P1) in this way, even if image distortion caused by a fisheye lens exists, the reference pixel (P0) and the actual pixel (P1) can be compared on the same or corresponding reference.

[0204] As illustrated in FIG. 7, when the autonomous flight drone (100) is located at the target separation distance (H), the number of pixels in the width direction of the marker band (200) measured on the center horizontal line (C3) can be calculated as actual pixels (P1) that are equal to or substantially identical to the reference pixels (P0). In this case, the control calculation unit (140) can determine that the autonomous flight drone (100) is maintaining the target separation distance (H), and the flight control unit (110) can maintain the current altitude.

[0205] On the other hand, when the autonomous flying drone (100) approaches the ceiling (11), the distance between the marker camera (130) and the marker band (200) decreases, so the number of pixels in the width direction of the marker band (200) captured through the fisheye lens may increase. In this case, the actual pixels calculated on the center horizontal line (C3) may appear as values ​​larger than the reference pixel (P0), such as P1a. The control calculation unit (140) may determine this state as the autonomous flying drone (100) being excessively close to the ceiling (11).

[0206] Conversely, when the autonomous flying drone (100) moves away from the ceiling (11), the distance between the marker camera (130) and the marker band (200) increases, so the number of pixels in the width direction of the marker band (200) captured through the fisheye lens may decrease. In this case, the actual pixels calculated on the center horizontal line (C3) may appear as values ​​smaller than the reference pixel (P0), such as P1b. The control calculation unit (140) can determine this state as the autonomous flying drone (100) having deviated below the target separation distance (H).

[0207] In this embodiment, the altitude correction value (Hc) can be calculated as the value obtained by subtracting the actual pixel (P1) from the reference pixel (P0), i.e., P0-P1. Alternatively, the altitude correction value (Hc) may be calculated as a value proportional to P0-P1. Here, the proportional value may be the value obtained by multiplying P0-P1 by a predetermined control gain, correction coefficient, or filtering coefficient.

[0208] If the actual pixel (P1) is greater than the reference pixel (P0), P0-P1 becomes a negative value. In this case, the control calculation unit (140) generates a negative altitude correction value (Hc), and the flight control unit (110) can perform descent control so that the autonomous flight drone (100) moves away from the ceiling (11). Conversely, if the actual pixel (P1) is smaller than the reference pixel (P0), P0-P1 becomes a positive value. In this case, the control calculation unit (140) generates a positive altitude correction value (Hc), and the flight control unit (110) can perform ascent control so that the autonomous flight drone (100) moves closer to the ceiling (11).

[0209] The flight control unit (110) can adjust the amount of control for ascent or descent control according to the magnitude of the altitude correction value (Hc). For example, if the absolute value of the altitude correction value (Hc) is large, the autonomous flight drone (100) is in a state of being significantly deviated from the target separation distance (H), so the flight control unit (110) can apply a relatively large amount of ascent or descent control. Conversely, if the absolute value of the altitude correction value (Hc) is small, a fine amount of control can be applied to reduce shaking of the autonomous flight drone (100) and restore the target separation distance (H).

[0210] Additionally, the flight control unit (110) can maintain the current altitude when the altitude correction value (Hc) is within a predetermined allowable range. The allowable range can be set by considering the image distortion of the fisheye lens, the detection error of the marker band (200), the noise of the identification image, the resolution of the marker camera (130), the minute vibration of the autonomous flight drone (100), or the control stability. Accordingly, unnecessary ascent or descent control can be suppressed in the range where the target separation distance (H) is substantially maintained, even if the reference pixel (P0) and the actual pixel (P1) do not completely match.

[0211] Additionally, the control calculation unit (140) can correct the actual pixel (P1) by taking into account the distortion characteristics of the fisheye lens. For example, the control calculation unit (140) can correct the actual pixel (P1) according to the projection model of the fisheye lens, the lens correction table, the distance from the center point (Cp1), or the area where the image portion of the marker band (200) is located within the identification image. Although such correction is not mandatory, it can help improve the accuracy of calculating the altitude correction value (Hc) even when the marker band (200) is captured away from the center of the identification image.

[0212] According to the present embodiment, an autonomous flying drone (100) uses a fisheye lens of an equidistant projection method to show a marker strip (200) marked on the ceiling (11) of an underground power conduit (10) with a wide field of view ( It can be taken with ). In addition, since it can be compared with the actual pixel (P1) using a reference pixel (P0) that reflects the projection characteristics of the fisheye lens, the altitude correction value (Hc) can be generated more accurately than when the standard reference pixel calculation formula for a center projection lens is applied as is.

[0213] Accordingly, the autonomous flight drone (100) can stably photograph the marker strip (200) and maintain the distance (H) from the ceiling (11) even in environments where the space is narrow and the distance from the ceiling (11) is short, such as inside an underground power conduit (10). In addition, in a close flight state, even a small change in the distance (H) can be sensitively manifested as a change in the field of view in the width direction of the marker strip (200) and a change in the actual pixel (P1), so the control calculation unit (140) can quickly detect even a small change in altitude of the autonomous flight drone (100). Therefore, the fisheye lens-based altitude control of this embodiment can be particularly advantageously applied in a relatively small underground power conduit where the ceiling height is low and the marker strip (200) extends mainly in a straight section. However, this is merely an example of an operating environment in which the fisheye lens embodiment is particularly suitable, and it does not exclude the application of the present invention to an underground power conduit that includes a curved section.

[0214] Next, we examine an embodiment of the present invention that generates path member information and safely returns an autonomous flight drone when a marker band or centerline is not identified in an identification image. This embodiment relates to a safe return control in case the marker band (200) or centerline (C0) is not identified temporarily or continuously during the process of the autonomous flight drone (100) autonomously flying inside an underground power conduit (10).

[0215] Referring to FIGS. 1 and 2, an autonomous flight drone (100) photographs a marker band (200) and a centerline (C0) marked on the ceiling (11) of an underground power conduit (10) using a marker camera (130), and a control calculation unit (140) can generate an altitude correction value (Hc), a left / right correction value (Lc), and a direction correction value (Dc) by analyzing the identification image captured by the marker camera (130). A flight control unit (110) can control the autonomous flight drone (100) to fly along the space below the marker band (200) using the correction values.

[0216] However, inside the underground power conduit (10), the marker strip (200) or centerline (C0) may not be properly identified in the identification image due to poor lighting, dust, condensation, leakage, contamination, damage to the marker strip (200), temporary obstruction by obstacles, shaking of the marker camera (130), changes in the aircraft attitude, image noise, or disturbances. Additionally, even if the autonomous flying drone (100) deviates significantly from the space below the marker strip (200) due to disturbances or control errors, the marker strip (200) or centerline (C0) may not be detected in the identification image.

[0217] In this embodiment, the control calculation unit (140) can repeatedly determine whether the marker band (200) or the center line (C0) is identifiable in the identification image. For example, the control calculation unit (140) can detect the color, brightness, saturation, width direction boundary line of the marker band (200), the color or linear pattern of the center line (C0) in the identification image, and determine whether they satisfy predetermined identification conditions. The identification conditions may include, for example, whether the number of pixels in the width direction of the marker band (200) exists within a reference range, whether the image portion of the center line (C0) is detected for a length greater than a certain length, whether the difference in color or brightness between the marker band (200) and the center line (C0) is greater than a reference value, or whether the detection reliability is greater than a predetermined reference value.

[0218] The control calculation unit (140) may not immediately perform return control when the marker band (200) or center line (C0) is temporarily not identified in the identification image, but may attempt re-identification for a predetermined time or a predetermined number of frames. For example, the control calculation unit (140) may determine a path loss state if the marker band (200) or center line (C0) is not identified in a series of consecutive identification images, or if the identification reliability is below a reference value and the condition persists for a predetermined time or longer. By applying conditions for multiple frames or a predetermined time in this way, the initiation of return control unnecessarily due to temporary image noise or instantaneous lighting changes can be reduced.

[0219] The control calculation unit (140) may generate path absence information when it is determined that a path is lost. The path absence information may be information indicating that a state has occurred in which a marker band (200) or a centerline (C0) is not identified in the identification image. The path absence information may include the time of path absence, the flight state immediately before the path absence occurred, the altitude correction value (Hc) immediately before the path absence occurred, the left / right correction value (Lc), the direction correction value (Dc), the estimation information of the cause of identification failure, the detection reliability of the marker band (200) or the centerline (C0), the attitude information of the autonomous flight drone (100), battery status information, or the immediate position information.

[0220] The control calculation unit (140) can transmit the generated path missing information to the flight control unit (110). When the flight control unit (110) receives the path missing information, it can determine that the autonomous flight drone (100) can no longer perform normal path-following flight based on the marker band (200) and centerline (C0), and can perform safety return control.

[0221] In this embodiment, when the flight control unit (110) receives path absence information, it can control the autonomous flight drone (100) to follow the previous flight path in the reverse direction. To this end, the storage unit (150) may store a flight log for a certain time or distance during the flight of the autonomous flight drone (100). The flight log may include control commands from the flight control unit (110), the direction of travel of the autonomous flight drone (100), altitude, speed, yaw angle, left / right movement amount, altitude correction value (Hc), left / right correction value (Lc), direction correction value (Dc), curve section entry information, location marker information, or time information.

[0222] The flight control unit (110) can control the autonomous flight drone (100) to follow the path it traveled immediately before by using the previous flight log stored in the storage unit (150) in the reverse direction. For example, the flight control unit (110) can control the autonomous flight drone (100) to retreat in a safe direction from the location where the path was missed by playing back the records of forward, left / right movement, ascent or descent, and yaw control for a certain period of time immediately before the path missing information occurred in reverse order, or by setting the position and attitude information corresponding to the previous flight path as the reverse target value.

[0223] Alternatively, when the flight control unit (110) receives path absence information, it can control the autonomous flight drone (100) to change its direction of travel by 180 degrees and return to the docking station (300). For example, even if the autonomous flight drone (100) is unable to identify the marker band (200) or the centerline (C0), it can reverse the direction of travel of the autonomous flight drone (100) using the flight direction and flight path information up to that point, and return to the docking station (300) using the path passed up to that point or the return path information stored in the storage unit (150).

[0224] At this time, the flight control unit (110) can perform temporary hovering before changing the direction of travel by 180 degrees so that the autonomous flight drone (100) does not rotate rapidly or approach a wall or cable hanger (13). In addition, the flight control unit (110) can perform return control after stabilizing the attitude of the autonomous flight drone (100) by using the obstacle detection value, barometric pressure sensor value, inertial sensor value, or gyro sensor value of the sensor unit (170) as an auxiliary tool.

[0225] The flight control unit (110) can continue to acquire identification images through the marker camera (130) even during safe return control. When the marker band (200) or centerline (C0) is identified again during the process of the autonomous flight drone (100) following the previous flight path in reverse or returning by rotating 180 degrees, the control calculation unit (140) can generate re-identification information. Upon receiving the re-identification information, the flight control unit (110) can realign the autonomous flight drone (100) to the marker band (200) and centerline (C0), and then continue the return flight or resume the inspection flight as necessary. However, if path missing information is generated, it may be preferable to set the drone to return to the docking station (300) with safety as the priority.

[0226] The storage unit (150) can store path absence information. Additionally, the storage unit (150) can store the time of occurrence of path absence information, the surveillance video at the time when path absence information occurred, or the identification video at the time when path absence information occurred. For example, among the surveillance videos captured by the surveillance camera (120), video corresponding to a certain time interval immediately before and after the occurrence of path absence information may be stored, and among the identification videos captured by the marker camera (130), video frames in which the marker band (200) or center line (C0) is not identified may also be stored.

[0227] In this way, by storing the surveillance video or identification video at the relevant time along with the path absence information, the cause of the path absence can be analyzed later. For example, the surveillance server (400) or the administrator can check the identification video at the time the path absence information occurred to determine whether the marker strip (200) is damaged, contaminated, has poor lighting, is obscured by an obstacle, has condensation, has leaked, or has a camera malfunction. Additionally, by checking the path absence information along with the surveillance video, it is possible to determine which section of the underground power conduit (10) the autonomous flight drone (100) lost its flight path.

[0228] When the autonomous flight drone (100) lands at the docking station (300), the communication unit (160) can transmit path omission information stored in the storage unit (150) to the docking station (300). At this time, the communication unit (160) can transmit not only the path omission information but also the time of occurrence of the path omission information, the surveillance video at the time the path omission information occurred, the identification video, the flight log, the battery status information, or the location marker information together to the docking station (300).

[0229] The docking station (300) can store path absence information and related data received from the autonomous flying drone (100) in its own storage unit or transmit them to the monitoring server (400) via the communication network (900). When the monitoring server (400) receives path absence information, it can analyze the location, time, and cause of the path absence and, if necessary, provide an inspection notification to the administrator terminal. For example, the monitoring server (400) can mark a section of the marker band (200) corresponding to the location where the path absence information occurred as a maintenance target section, or request an inspection of the contamination, damage, or lighting condition of the marker band (200) in that section.

[0230] According to the present embodiment, the autonomous flight drone (100) can return to the docking station (300) by generating path missing information and following the previous flight path in reverse or changing direction by 180 degrees, rather than continuing to attempt flight and increasing the risk of collision even when the marker band (200) or centerline (C0) cannot be identified in the identification image. Accordingly, the autonomous flight drone (100) can be safely recovered even in the event of poor lighting, damaged markers, image noise, disturbance, or control error.

[0231] In addition, according to the present embodiment, the time of occurrence, surveillance video, or identification video is stored along with path absence information, and can be transmitted after the autonomous flight drone (100) lands at the docking station (300). Therefore, even if real-time communication is difficult during flight due to poor communication conditions inside the underground power conduit (10), post-analysis of the path absence situation is possible. Thus, the operational stability of the autonomous flight drone (100) can be increased, and the sections requiring maintenance of the marker band (200) or centerline (C0) can be quickly identified.

[0232] Next, we examine an embodiment in which the inspection flight state or return flight state of an autonomous flight drone is determined using the arrangement relationship between the first band area and the second band area of ​​the marker band in the present invention. This embodiment is intended to distinguish whether the autonomous flight drone (100) is currently heading from the inspection start point to the inspection end point, or returning from the inspection end point to the inspection start point, even when the autonomous flight drone (100) is identifying the marker band (200) and the center line (C0).

[0233] Referring to FIGS. 1 and 2, the marker band (200) may include a first band area (201) and a second band area (202) that are separated from each other based on the center line (C0). Here, the first band area (201) may be an area located to the left of the center line (C0) based on the direction from the inspection start point to the inspection end point, and the second band area (202) may be an area located to the right of the center line (C0) based on the direction from the inspection start point to the inspection end point.

[0234] The first band area (201) and the second band area (202) may have distinct colors, brightness, or saturation. For example, the first band area (201) may have a relatively bright color or high brightness, and the second band area (202) may have a relatively dark color or low brightness. Alternatively, the first band area (201) and the second band area (202) may have different colors, different saturation, different reflectances, or different patterns. Accordingly, the control calculation unit (140) can distinguish and recognize areas located on both the left and right sides relative to the center line (C0) in the identification image captured by the marker camera (130).

[0235] In this embodiment, the reason for forming the first band area (201) and the second band area (202) distinctly is not merely to determine whether the autonomous flight drone (100) is located in the space below the center line (C0), but to determine which direction the autonomous flight drone (100) is following the center line (C0). Since the center line (C0) itself is a linear mark extending in the longitudinal direction, it may be difficult to distinguish whether the autonomous flight drone (100) is heading from the inspection start point to the inspection end point, or from the inspection end point to the inspection start point, based solely on the center line (C0). Accordingly, in this embodiment, the band areas on the left and right sides are formed differently based on the center line (C0), so that the flight direction can be determined using the left-right arrangement relationship in the identification image.

[0236] The storage unit (150) may store the normal arrangement relationship of the first band area (201), centerline (C0), and second band area (202) that must be identified in the identification image when the autonomous flight drone (100) flies from the inspection start point to the inspection end point. The normal arrangement relationship may be, for example, a relationship in which the first band area (201), centerline (C0), and second band area (202) appear sequentially from left to right in the identification image. However, the normal arrangement relationship may be stored in a pre-corrected form according to the installation direction of the marker camera (130), the image coordinate system, whether image inversion processing is performed, or the reference coordinate system of the drone.

[0237] When the control calculation unit (140) receives an identification image from the marker camera (130), it can detect the marker band (200), the center line (C0), the first band area (201), and the second band area (202) in the identification image. For example, the control calculation unit (140) can detect the center line (C0) by using the difference between the color, brightness, or saturation of the center line (C0) and the color, brightness, or saturation of the marker band (200), and can identify the first band area (201) and the second band area (202) by comparing the color, brightness, or saturation of the left area and the right area based on the center line (C0).

[0238] The control calculation unit (140) can calculate the current arrangement relationship of the first band area (201), center line (C0), and second band area (202) identified in the identification image. The current arrangement relationship may be information indicating the order in which the first band area (201), center line (C0), and second band area (202) appear based on the left-right direction of the identification image. For example, the current arrangement relationship may be calculated in the order of “first band area (201)-center line (C0)-second band area (202)” or “second band area (202)-center line (C0)-first band area (201)”.

[0239] The control calculation unit (140) can compare the calculated current arrangement relationship with the normal arrangement relationship stored in the storage unit (150). If the current arrangement relationship matches the normal arrangement relationship, the control calculation unit (140) can determine that the autonomous flight drone (100) is in an inspection flight state heading from the inspection start point to the inspection end point. That is, it can determine that the autonomous flight drone (100) is following the marker band (200) in the normal inspection direction.

[0240] Conversely, if the current arrangement relationship is identified as being reversed left and right from the normal arrangement relationship, the control calculation unit (140) can determine that the autonomous flight drone (100) is in a return flight state heading from the inspection end point to the inspection start point. For example, if the normal arrangement relationship is stored such that the first band area (201) appears on the left side of the identification image and the second band area (202) appears on the right side during the inspection flight state, in the return flight state, the autonomous flight drone (100) faces the same marker band (200) by reversing its direction of travel, so the left and right positions of the first band area (201) and the second band area (202) may appear reversed in the identification image. The control calculation unit (140) can determine that the autonomous flight drone (100) is in a return flight state by using this reversed left and right arrangement relationship.

[0241] In this embodiment, the result of determining the inspection flight state or the return flight state can be used for flight control of the flight control unit (110). For example, if the flight control unit (110) determines that the inspection flight state is the inspection flight state, it can maintain forward flight from the inspection start point toward the inspection end point and continue to perform surveillance video recording using the surveillance camera (120). On the other hand, if the return flight state is determined, it can perform a return flight from the inspection end point toward the inspection start point and control the flight to move along the marker band (200) and centerline (C0) until the start mark is identified again or the docking station return path information is executed.

[0242] Additionally, the result of determining the inspection flight state or the return flight state can also be used for curve section control. Even with the same curve marker (210), the interpretation of the start or end point of the curve section may differ depending on whether the autonomous flight drone (100) is in an inspection flight state or a return flight state. Accordingly, the control calculation unit (140) can determine the current flight state through the current arrangement relationship of the first band area (201) and the second band area (202), and use this together with the identification ID, radius of curvature (R), and bending direction information of the curve marker (210) to generate curve section entry information or curve section end information.

[0243] Additionally, the control calculation unit (140) may determine that an identification error has occurred in the marker band (200) or centerline (C0) if the current arrangement relationship does not match the normal arrangement relationship and does not match the left-right inversion relationship of the normal arrangement relationship. For example, if the first band area (201) or the second band area (202) is not properly identified due to contamination, damage, shadows, condensation, or poor lighting, the current arrangement relationship may be calculated ambiguously. In this case, the control calculation unit (140) may perform re-identification for a certain period of time, or if the identification reliability is below a reference value, request the flight control unit (110) to perform deceleration, hovering, realignment, or safety return control.

[0244] Additionally, the storage unit (150) may store reference values ​​or allowable ranges regarding the color, brightness, or saturation of the first band area (201), the center line (C0), and the second band area (202). The control calculation unit (140) may determine each area by comparing the color value, brightness value, or saturation value of each area detected in the identification image with the reference values ​​stored in the storage unit (150). According to an embodiment, the control calculation unit (140) may distinguish the first band area (201) and the second band area (202) using relative brightness differences, color order, or contrast values ​​between areas to correspond to changes in illumination, camera exposure, or ceiling surface reflectance.

[0245] According to the present embodiment, in addition to controlling altitude, left / right position and direction of travel using a marker band (200) and a centerline (C0), the autonomous flight drone (100) can also determine the current flight direction using the arrangement relationship between the first band area (201) and the second band area (202). Accordingly, even when the autonomous flight drone (100) rotates 180 degrees at the inspection end point to perform a return flight, the current flight state can be reliably distinguished using the same marker band (200).

[0246] In addition, according to the present embodiment, even if the direction of travel of the autonomous flight drone (100) is temporarily confused due to disturbances, sudden yaw rotation, control errors, or path realignment, the inspection flight state and return flight state can be quickly determined by checking the current arrangement relationship of the first band area (201), centerline (C0), and second band area (202) in the identification image. Therefore, the autonomous flight drone (100) is prevented from flying in the wrong direction, and more stable inspection flight and return flight can be performed inside the underground power conduit (10).

[0247] Next, we examine an embodiment of the present invention that stores a location identification ID or distance traveled as metadata in a surveillance video using a location marker marked on a marker strip. This embodiment is intended to enable subsequent identification of the location within the underground power tunnel where a surveillance video captured by a surveillance camera was taken, even when identical or similar cables, cable hangers, and auxiliary facilities are repeatedly arranged inside the underground power tunnel.

[0248] Referring to FIGS. 1 and 2, a marker strip (200) extending from the inspection start point to the inspection end point may be marked on the ceiling (11) of the underground power conduit (10), and a center line (C0) may be marked at the center of the width direction of the marker strip (200). Additionally, a position marker (230) may be further marked on the marker strip (200). The position markers (230) may be placed at predetermined intervals from the inspection start point to the inspection end point.

[0249] The position marker (230) may be placed within the area of ​​the marker band (200). For example, the position marker (230) may be displayed in the first band area (201) or the second band area (202) so as not to overlap with the center line (C0), and may also be displayed at a location that can be identified together with a part of the center line (C0) according to the embodiment. It is preferable that the position marker (230) have a shape, color, brightness, saturation, or pattern that can be distinguished from the marker band (200), the center line (C0), the curve marker (210), or the start and end marks.

[0250] The location marker (230) may be formed from characters, numbers, geometric patterns, color patterns, barcodes, QR codes, 2D codes, or a combination thereof. For example, the location marker (230) may include a location identification ID such as “P001” or “P002”, or may include a distance traveled from the starting point of the inspection to the location where the location marker (230) is displayed, such as “50m” or “100m”. Additionally, the location marker (230) may include both the location identification ID and the distance traveled.

[0251] The interval at which the location markers (230) are displayed can be set according to the length of the underground power conduit (10), inspection precision, the need to specify the location of the surveillance video, the resolution of the marker camera (130), or the analysis unit of the surveillance server (400). For example, the location markers (230) can be displayed at intervals of 5m, 10m, 20m, or 50m, and may be displayed at shorter intervals in curved sections, junction boxes, branching sections, cable connection sections, drainage facilities, or sections of high importance for inspection.

[0252] The marker camera (130) can capture identification images of the marker band (200), centerline (C0), and position marker (230) while the autonomous flying drone (100) is flying. The control calculation unit (140) receives the identification images from the marker camera (130) and can determine whether the position marker (230) is identified in the identification images. The control calculation unit (140) can analyze the shape, pattern, character, number, or code information of the position marker (230) to extract the position identification ID or distance traveled included in the position marker (230).

[0253] For example, if the location marker (230) is formed as a QR code or a two-dimensional code, the control calculation unit (140) can detect the corresponding code area in the identification image and extract a location identification ID or distance traveled through code decoding. If the location marker (230) is formed as a number or a character, the control calculation unit (140) can extract a location identification ID or distance traveled using a character recognition algorithm. If the location marker (230) is formed as a color pattern or a shape pattern, the control calculation unit (140) can calculate a location identification ID or distance traveled corresponding to the location marker (230) by referring to a pattern-location correspondence table stored in the storage unit (150).

[0254] When a location marker (230) is identified, the control calculation unit (140) can transmit the extracted location identification ID or distance traveled to the storage unit (150). The storage unit (150) can store the location identification ID or distance traveled as metadata of the surveillance video along with the surveillance video captured by the surveillance camera (120). Here, the metadata may be stored in the form of header information of the surveillance video file, additional information per video frame, tag information per video segment, a separate index file, or a database record.

[0255] The metadata of the surveillance video may include not only the location identification ID or distance traveled, but also the time of shooting, flight status, whether it is an inspection flight or a return flight, altitude correction value (Hc), left / right correction value (Lc), direction correction value (Dc), battery status, curve section information, path absence information, or the identification reliability of the location marker (230). It is desirable that such metadata be stored so as to correspond temporally with the surveillance video.

[0256] For example, at the time when the autonomous flying drone (100) identifies the location marker (230), the location identification ID or distance traveled of the corresponding location marker (230) may be assigned to the video frame or video segment captured by the surveillance camera (120). Additionally, relative location information based on the location identification ID or distance traveled of the previous location marker (230) may be assigned to the video segment from when the location marker (230) is identified until the next location marker (230) is identified.

[0257] According to an embodiment, the control calculation unit (140) may interpolate the intermediate position of the surveillance video using the flight speed, flight time, estimated distance of the autonomous flight drone (100), or the interval between the position markers between two continuously identified position markers (230). For example, if the first position marker is at a position 100m from the inspection start point and the next second position marker is at a position 110m, the control calculation unit (140) may assign interpolated distance information such as 101m, 102m, etc., to each frame of the surveillance video captured between the two position markers according to the time of capture or the estimated distance of movement.

[0258] Additionally, when the autonomous flight drone (100) is in a return flight state, the control calculation unit (140) can determine that the current flight state is a return flight state by using the arrangement relationship of the first band area (201), the centerline (C0), and the second band area (202) described in claim 9. In this case, the distance traveled included in the position marker (230) is maintained as the distance from the inspection start point, but the actual direction of travel of the autonomous flight drone (100) can be stored together as meta-information that it is the direction from the inspection end point to the inspection start point. Accordingly, the same position marker (230) can be managed by distinguishing between cases where it is identified during an inspection flight and cases where it is identified during a return flight.

[0259] The storage unit (150) stores the surveillance video and the location identification ID or distance traveled in correspondence with each other, thereby making it easy to identify the location when an abnormal sign is detected in the surveillance video. For example, if overheating, discoloration, damage, sagging, connection abnormality, deformation or breakage of the cable hanger (13), leakage, flooding, condensation, inflow of foreign matter, or equipment damage is confirmed in the surveillance video, the location of the abnormality can be calculated using the location identification ID or distance traveled corresponding to the video frame or video section.

[0260] When the autonomous flying drone (100) finishes its inspection flight and lands at the docking station (300), the communication unit (160) can transmit the surveillance video stored in the storage unit (150) and the corresponding metadata to the docking station (300). The docking station (300) can store the received surveillance video and metadata internally or transmit them to the surveillance server (400) via the communication network (900).

[0261] The monitoring server (400) can diagnose whether there is an abnormality in the underground power conduit (10) and the underground power distribution line (14) by analyzing the monitoring video and metadata received from the docking station (300). At this time, if an abnormal sign is detected in the monitoring video, the monitoring server (400) can display the location identification ID or distance traveled corresponding to the video frame or video section. For example, the monitoring server (400) can output the location where the abnormality occurred, such as “near location identification ID P023”, “230m from the inspection start point”, or “the section between P023 and P024”.

[0262] Additionally, the monitoring server (400) can manage the inspection history of the underground power conduit (10) by section using a location identification ID or travel distance. For example, the monitoring server (400) can analyze the degree of deformation of the cable hanger (13), the extent of leakage, the condition of the cable surface, or the amount of change in temperature distribution by comparing past monitoring images and current monitoring images corresponding to the same location identification ID. Accordingly, it can determine whether the abnormal condition at a specific location is worsening over time.

[0263] In addition, metadata based on location markers (230) can also be utilized for dispatching maintenance workers to the site. When the monitoring server (400) provides the location of an anomaly as a location identification ID or travel distance, the worker can move the corresponding travel distance from the starting point of the inspection of the underground power conduit (10) or find the location marker (230) displaying the location identification ID to reach the location of the anomaly. Accordingly, the problem of difficulty in identifying the location using only video due to the repetitive cable and cable hanger structures can be reduced.

[0264] In this embodiment, the position marker (230) may be used together with the curve marker (210), the start mark, or the end mark. For example, a start mark may be displayed at the inspection start point, an end mark at the inspection end point, and a position marker (230) may be displayed at regular intervals between them. Additionally, a curve marker (210) may be displayed at the start and end points of the curve section. In this case, the control calculation unit (140) can distinguish the start mark, end mark, curve marker (210), and position marker (230) from each other by using the shape, code system, or type of identification ID of each marker.

[0265] According to the present embodiment, the autonomous flying drone (100) can store the location identification ID or distance traveled included in the location marker (230) along with the surveillance video captured by the surveillance camera (120) as metadata. Accordingly, even when identical or similar structures are repeatedly placed inside the underground power conduit (10), the location where the surveillance video was captured can be accurately identified.

[0266] In addition, according to the present embodiment, the monitoring server (400) can display the location of an abnormal occurrence by using the results of the monitoring video analysis together with the location identification ID or travel distance information. Therefore, when an abnormal sign is detected in the underground power distribution line (14), a maintenance worker can quickly locate the corresponding location and improve the efficiency of subsequent maintenance work.

[0267] Although the present invention has been described with various examples as stated above, the present invention is not necessarily limited to these examples and may be implemented with various modifications within the scope of the technical concept of the present invention. Accordingly, the examples disclosed in the present invention are intended to explain, not limit, the technical concept of the present invention, and the scope of the technical concept of the present invention is not limited by these examples. The scope of protection of the present invention shall be interpreted by the claims below, and all technical concepts within an equivalent scope shall be interpreted as being included within the scope of rights of the present invention. Industrial applicability

[0268] The present invention can be utilized in the fields of unmanned inspection, video surveillance, equipment maintenance, detection of abnormal signs, and remote management of underground power conduits, underground utility tunnels, underground power distribution lines, power cable housing facilities, or similar underground infrastructure. In particular, the present invention has industrial applicability as it can be applied to a system that acquires surveillance footage using an autonomous drone inside an underground power conduit where direct entry by workers is difficult or dangerous, and diagnoses abnormalities in underground power distribution lines in conjunction with a docking station and a monitoring server. Explanation of the symbols

[0269] 10: Underground power conduit 11: Ceiling 12: Floor 13: Cable hanger 14: Underground power distribution line 100: Autonomous flying drone 100a: Autonomous flight drone close to the ceiling 100b: Autonomous flying drone away from the ceiling 110: Flight Control Unit 120: Surveillance Camera 130: Marker Camera 131: Marker Gimbal 140: Control calculation unit 150: Storage unit 160: Communication unit 170: Sensor unit 200: Marker band 201: First band area 202: Second band area 210: Curve marker 230: Location Marker 300: Docking Station 400: Monitoring server 900: Communication network B1: First boundary line B2: Second boundary line C0: Center line C1: Center vertical line C3: Center horizontal line Cp1: Center point Cp2: Centerline midpoint D: Width of marker strip d1: Left / Right deviation distance Dc: Direction correction value f: Focal length H: Distance Ha: Distance when close to the ceiling Hb: Distance from the ceiling Hc: Altitude correction value Lc: Left / Right correction value P0: Reference pixel P1: Actual pixel P1a: Actual pixel in a state close to the ceiling P1b: Actual pixel distanced from the ceiling P2: Number of relative pixels in the left-right direction Pw: Total number of pixels in the horizontal direction of the image sensor R: Radius of curvature W: Horizontal length of the image sensor θ: Intersection angle x, y: Relative coordinates : Angle of view of a marker camera

Claims

Claim 1 In an autonomous flying drone for monitoring underground power distribution lines capable of autonomously flying inside an underground power conduit and capturing surveillance images of the underground power conduit and underground power distribution lines, the ceiling of the underground power conduit is provided with a marker band formed in the shape of a band having a certain width (D) and having a color contrasting with the color of the ceiling surface of the underground power conduit, and extending from the inspection start point to the inspection end point of the underground power conduit; the marker band is provided with a center line (C0) located at the center in the width direction of the marker band and extending along the length direction of the marker band, and the center line (C0) has a color, brightness, or saturation that is distinguishable from the color of the marker band; the autonomous flying drone includes a flight control unit, a surveillance camera, a marker camera, a marker gimbal, a control calculation unit, a storage unit, and a communication unit; the surveillance camera captures surveillance images of the underground power conduit and the underground power distribution lines while the autonomous flying drone is flying and stores them in the storage unit; and the marker camera is While the autonomous flying drone is flying, it captures identification images of the marker strip and the centerline (C0) and transmits them in real time to the control computing unit; the marker gimbal controls the attitude of the marker camera so that the lens optical axis of the marker camera is vertically oriented toward the ceiling of the underground power conduit; the storage unit stores reference pixels (P0), which are the number of pixels in the width direction of the marker strip captured when the marker camera captures the identification image at a certain distance (H) from the ceiling of the underground power conduit; the communication unit can connect to the docking station via a short-range communication network when the autonomous flying drone lands at a docking station installed in the underground power conduit and transmit the surveillance images stored in the storage unit to the docking station; the control computing unit analyzes the marker strip and the centerline (C0) captured in the identification image to generate an altitude correction value (Hc), a left-right correction value (Lc), and a direction correction value (Dc), and then transmits them to the flight control unit.The flight control unit controls the autonomous flight drone to take off from the docking station, fly autonomously along the space below the marker strip from the inspection start point to the inspection end point, and then return to the docking station, wherein: - altitude control is performed using the altitude correction value (Hc) so that the autonomous flight drone can fly forward while maintaining the separation distance (H) from the ceiling of the underground power conduit; - horizontal movement control is performed using the left-right correction value (Lc) so that the autonomous flight drone can fly forward along the space below the centerline (C0); - horizontal angle (yaw) control is performed using the direction correction value (Dc) so that the autonomous flight drone can fly forward while maintaining its direction of travel parallel to the length direction of the centerline (C0); the reference pixel (P0) is calculated using the width (D) of the marker strip, the separation distance (H), and the specifications of the marker camera, and the marker strip captured in the identification image An autonomous flying drone for monitoring an underground power distribution line, characterized in that the value calculated from the number of pixels in the width direction of the image is the actual pixel (P1), the altitude correction value (Hc) is generated based on the difference between the actual pixel (P1) and the reference pixel (P0), the left-right correction value (Lc) is generated based on the relative position in the left-right direction between the center point (Cp1) located at the exact center of the identification image and the center line midpoint (Cp2), the direction correction value (Dc) is generated based on the difference in the intersection angle or slope between the center line (C0) and the center vertical line (C1) captured in the identification image, the center line midpoint (Cp2) is the point that bisects the image portion of the center line (C0) captured in the identification image in the length direction, and the center vertical line (C1) is a line segment connecting the upper side midpoint and the lower side midpoint of the identification image. Claim 2 In claim 1, the specifications of the marker camera include a focal length (f), a horizontal length (W) of the image sensor, and a total number of horizontal pixels (Pw) of the image sensor, and the reference pixel (P0) is calculated according to the following formula using the width (D) of the marker band, the separation distance (H), the focal length (f), the horizontal length (W) of the image sensor, and the total number of horizontal pixels (Pw) of the image sensor. The above altitude correction value (Hc) is a value obtained by subtracting the actual pixel (P1) from the reference pixel (P0) (P0-P1), or a value proportional to the subtracted value (P0-P1); the flight control unit, wherein - maintains the current altitude of the autonomous flight drone when the altitude correction value (Hc) is within a predetermined allowable range, - performs descent control so that the autonomous flight drone moves away from the ceiling of the underground power conduit when the altitude correction value (Hc) is a negative value outside the allowable range, - performs ascent control so that the autonomous flight drone moves closer to the ceiling of the underground power conduit when the altitude correction value (Hc) is a positive value outside the allowable range, and - increases the amount of control for ascent control or descent control as the absolute value of the altitude correction value (Hc) increases. Claim 3 In claim 1, the control calculation unit calculates the number of left-right relative pixels (P2) between the center point (Cp1) and the center line midpoint (Cp2), and calculates the left-right deviation distance (d1) of the autonomous flight drone according to the following formula using the width (D) of the marker band, the actual pixel (P1), and the number of left-right relative pixels (P2). - An autonomous flight drone for monitoring an underground power distribution line, characterized by generating a left-right correction value (Lc) including a deviation direction indicating whether the left-right deviation distance (d1) and the centerline midpoint (Cp2) are located to the left or right relative to the center point (Cp1), and transmitting this to the flight control unit, and the flight control unit performing left-right horizontal movement control of the autonomous flight drone using the left-right correction value (Lc) so that the centerline midpoint (Cp2) in the identification image coincides with the center point (Cp1). Claim 4 An autonomous flight drone for monitoring an underground power distribution line according to claim 1, wherein the direction correction value (Dc) is the intersection angle (θ) formed by the image portion of the center line (C0) captured in the identification image with respect to the center vertical line (C1), or a value proportional to the intersection angle (θ), having a positive value when the image portion of the center line (C0) is tilted clockwise and a negative value when it is tilted counterclockwise; the flight control unit maintains the current flight angle of the autonomous flight drone when the direction correction value (Dc) is within a predetermined allowable range, and performs horizontal angle (yaw) control for the autonomous flight drone so that the intersection angle (θ) approaches 0 when the direction correction value (Dc) exceeds the allowable range, wherein the amount of control for the horizontal angle (yaw) control increases as the absolute value of the direction correction value (Dc) increases, and the control calculation unit calculates the intersection angle (θ) as shown in the following equation. (x, y are relative coordinates for an arbitrary point other than the intersection point in the image portion of the center line (C0), where the intersection point of the image portion of the center line (C0) and the center vertical line (C1) is set as the coordinate origin (0,0).) Claim 5 In claim 1, the underground power conduit includes a curved section, and at each of the starting and ending positions of the curved section, a curve marker is displayed that is placed within the area of ​​the marker band; each of the curve markers includes an identification ID, a radius of curvature (R) of the curved section, and a bending direction when entering the curved section based on the corresponding curve marker; when the curve marker is identified in the identification image, the control calculation unit generates curved section entry information or curved section end information using the identification ID, the radius of curvature (R), and the bending direction included in the curve marker; and when the flight control unit receives the curved section entry information, it performs horizontal angle (yaw) control based on the direction correction value (Dc) so that the tangential direction or local length direction of the portion of the image part of the centerline (C0) located in front of the autonomous flight drone's direction of travel remains parallel to the center vertical line (C1), and the autonomous flight An autonomous flight drone for monitoring an underground power distribution line, characterized by correcting the direction of travel of the drone, and the flight control unit, upon receiving information on the end of the curved section, terminating the correction of the direction of travel corresponding to the curved section and controlling the drone to fly forward along the length direction of the centerline (C0). Claim 6 In claim 5, the control calculation unit is characterized by: - ​​counting the identification order of the curve markers when the curve markers are sequentially identified in the identification image while the autonomous flight drone is flying from the inspection start point to the inspection end point; - determining the curve marker as a curve section start marker displayed at the starting point of the curve section when the identification order is odd, - determining the curve marker as a curve section end marker displayed at the end point of the curve section when the identification order is even, - generating curve section entry information including the identification ID, the radius of curvature (R), and the bending direction when the curve section start marker is identified, and - generating curve section end information to terminate the correction of the direction of travel corresponding to the curve section when the curve section end marker is identified. Claim 7 In claim 1, the marker camera includes a fisheye lens of an equidistant projection method, and the specifications of the marker camera include the focal length (f) of the fisheye lens, the horizontal length (W) of the image sensor, and the total number of horizontal pixels (Pw) of the image sensor, and the reference pixel (P0) is calculated according to the following formula using the width (D) of the marker band, the separation distance (H), the focal length (f) of the fisheye lens, the horizontal length (W) of the image sensor, and the total number of horizontal pixels (Pw) of the image sensor. The above control calculation unit calculates the number of pixels in the width direction of the marker band image captured in the identification image as the actual pixel (P1), and the altitude correction value (Hc) is a value obtained by subtracting the actual pixel (P1) from the reference pixel (P0) (P0-P1) or a value proportional to the subtracted value (P0-P1), and the flight control unit performs altitude control using the altitude correction value (Hc) so that the autonomous flight drone maintains the separation distance (H) from the ceiling of the underground power conduit, characterized by an autonomous flight drone for monitoring an underground power distribution line. Claim 8 An autonomous flight drone for monitoring an underground power distribution line according to claim 1, wherein the control calculation unit generates path absence information and transmits it to the flight control unit when the marker band or the centerline (C0) is not identified in the identification image, and the flight control unit, upon receiving the path absence information, controls the autonomous flight drone to follow the previous flight path in reverse or change its direction of travel by 180 degrees to return to the docking station, the storage unit stores the path absence information, the time of occurrence of the path absence information, the surveillance image at the time the path absence information occurred, or the identification image at the time the path absence information occurred, and the communication unit transmits the path absence information stored in the storage unit to the docking station when the autonomous flight drone lands at the docking station. Claim 9 In claim 1, the marker band includes a first band area located to the left of the centerline (C0) and a second band area located to the right of the centerline (C0), based on the direction from the inspection start point to the inspection end point, wherein the first band area and the second band area have distinct colors, brightness, or saturation, and the storage unit stores the normal arrangement relationship of the first band area, the centerline (C0), and the second band area that must be identified in the identification image when the autonomous flight drone flies from the inspection start point to the inspection end point, and the control calculation unit: - compares the current arrangement relationship of the first band area, the centerline (C0), and the second band area identified in the identification image with the normal arrangement relationship, - determines that if the current arrangement relationship matches the normal arrangement relationship, the autonomous flight drone is in an inspection flight state heading from the inspection start point to the inspection end point, and - the current arrangement relationship is left and right with the normal arrangement relationship An autonomous flight drone for monitoring an underground power distribution line, characterized by determining that, in the opposite case, the autonomous flight drone is in a return flight state heading from the inspection end point to the inspection start point. Claim 10 An autonomous flying drone for monitoring an underground power distribution line according to claim 1, wherein the marker strip further displays position markers at predetermined intervals from the inspection start point to the inspection end point, wherein the position markers include a position identification ID for identifying the location where the position markers are displayed within the underground power conduit or a distance traveled from the inspection start point to the location where the position markers are displayed, wherein the control calculation unit extracts the position identification ID or the distance traveled included in the position markers when the position markers are identified in the identification image, and the storage unit stores the position identification ID or the distance traveled as metadata of the surveillance image together with the surveillance image captured by the surveillance camera.

Citation Information

Patent Citations

  • Mobile-robot-carried camera position calibration method

    CN103020957A

  • Measurement method for determining the position of an aircraft or motor vehicle relative to the side of a runway or road from an image obtained using a camera that includes at least one edge of the runway or road

    DE10305993A1

  • Inspection device for duct inner wall and computer program

    JP2021011266A

  • Unmanned Transportation System based on linetracing of Ceiling type for interlocking with unmanned drone and Unmanned transportation service method using the same

    KR1020160056671A

  • Tracking Driving System for Logistics Transfer Robot

    KR1020210021746A