KR101836229B1 - Obstacle recognition using camera and infrared sensor and movement method for robots to climb up walls - Google Patents

Abstract

The present invention relates to a moving method of a wall-mounted mobile robot using a camera and an infrared sensor according to an obstacle recognition. It uses an infrared sensor provided in the robot to determine whether or not there is an obstacle ahead, The width of the obstacle is measured through image processing of the obstacle taken by the camera to avoid the case where the width of the obstacle is narrower than the set value. If the width of the obstacle is larger than the set value, The present invention provides an effect of overcoming the limitations of the conventional wall-mounted mobile robot and improving the usability and reliability of the wall-mounted mobile robot.

Description

Technical Field [0001] The present invention relates to an obstacle recognition method using a camera and an infrared sensor, and more particularly,

BACKGROUND OF THE INVENTION 1. Field of the Invention [0002] The present invention relates to an obstacle recognizing method and a coping method of a wall-moving robot moving along a wall with a multi-joint leg.

Work in extreme environments such as cleaning exterior walls of large structures, welding of large ships, and safety inspection of various industrial tanks, threatens workers' safety and costs a lot. A variety of wall-mounted robots are being developed to replace people in these tasks. The wall-mounted robot has a higher complexity than the general robot in the way of attaching the robot to the wall against the gravity and moving in the attached state.

There are five types of wall attachment methods for wall-mounted robots. First, in the electrostatic method, the wall attachment is implemented by alternately flowing (-) and (+) electricity and pulling each other like a magnet. However, this method can not be applied to walls that do not have electricity. The second method using the lizard pad with fine hair is difficult to apply to the outer wall of the building because there is a problem that dust or small sands stick to the fur and it does not contact the wall. Third, the method using the magnet can not be used in a structure other than the magnetic body, and the method using the fourth rake has a limitation in that a device for raising the wall can be installed in advance. Finally, in the vacuum method, it is possible to generate a large force by atmospheric pressure and vacuum, but it is difficult to keep the vacuum state inside the robot when moving from the wall to the ceiling, and there is a problem that power consumption is large by using an air compressor.

Methods of moving the robot attached to the wall include a method using a wheel, an infinite orbit shape, and a method using a plurality of legs. In the case of the wheel-based method and the infinite-track method, it is impossible to cope with obstacles in the movement path. In the method using multiple legs, there is a problem that the moving speed is slow because it needs to be detached and moved.

In addition, the conventional wall-mounted mobile robot has a problem in that it can not move freely on the wall surface where the obstacle exists, by using a method of unconditionally avoiding the obstacle when the obstacle is encountered or by instructing the user to cope with the obstacle.

The contents of the background art described above are technical information that the inventor of the present application holds for the derivation of the present invention or acquired in the derivation process of the present invention and is a known technology disclosed to the general public prior to the filing of the present invention I can not.

Korean Patent No. 10-1279501 Korean Patent Publication No. 10-2016-0026436

SUMMARY OF THE INVENTION The present invention has been conceived to solve the problems described above, and it is an object of the present invention to overcome the limitations of the conventional wall-mounted mobile robot by constructing a camera and an infrared sensor so as to recognize obstacles, The object of the present invention is to provide a moving method according to obstacle recognition of a wall mover using a camera and an infrared sensor capable of improving usability and reliability.

According to an embodiment of the present invention, there is provided a method of moving a wall-mounted robot using a camera and an infrared sensor according to the present invention for realizing the above-mentioned object, the method comprising: determining whether an obstacle is present in front of the obstacle, A first step of calculating a distance to the obstacle; If it is determined that there is an obstacle in the first step, the width of the obstacle is measured through image processing of the obstacle taken by the camera, thereby avoiding the case where the width of the obstacle is narrower than the set value. If the width of the obstacle is larger than the set value, And a second step of moving the robot in such a manner that the robot moves.

The method of calculating the distance in the first step includes: reading the distance value of n times, sorting the values, removing a value having a difference equal to or larger than the tolerance range e specified by the user from the median value as noise, It is preferable to determine the final distance by obtaining an average of the values that have not been obtained.

In the first step, noise is removed and the final distance is calculated.

이라고 하고,(1) read data by the desired number n and collect it

And,

는 상기 집합

을 정렬한 값들의 집합으로 설정할 때, (2) set

Lt; RTI ID =

To a set of sorted values,

에서 사용자가 설정한 오차 허용 범위 e의 ±값을 통해 값을 정제하여 아래 [수학식 1]과 같이 집합

를 산출하고,(3) set

, The value is refined through the ± value of the error tolerance range e set by the user,

Lt; / RTI >

에 속한 모든 원소들의 평균은

라고 정의하면, (4)

The average of all elements in

By definition,

(5) The standard deviation s is calculated by the following equation (2)

(6) The 95% confidence interval is calculated through the following equation (3) through the standard deviation s,

는 아래의 [수학식 4]를 이용하여 집합

의 모든 원소 중

함수에 참인 값들의 집합으로 정의하고, (7) set

(4) using the following equation

Of all elements of

Define the function as a set of values that are true,

는 집합

의 평균으로 정의하여 아래의 [수학식 5]를 통해 산출하는 것을 특징으로 하는 카메라와 적외선 센서를 이용한 벽면 이동 로봇의 장애물 인식에 따른 이동 방법.(8) Distance value to be obtained

Is a set

And calculating the distance through the following formula (5): " (5) "

[Equation 1]

&Quot; (2) "

&Quot; (3) "

&Quot; (4) "

&Quot; (5) "

In the second step, when the inclination of the obstacle is not 0, it is preferable to set the robot so as to rotate by the inclination.

The robot includes a body; Four legs each connected to the body and having a six-joint motion structure by respective motors; An adsorption mechanism for forming vacuum pressure on a vacuum pad provided at an end of each of the legs to generate an attraction force for attaching each leg to a wall surface; And a control mechanism for controlling the motors and the adsorption mechanism provided on the respective legs.

The legs include six servomotors capable of six-joint operation, wherein the first servo motor is configured to rotate the leg about the Y axis with respect to the body, And the third and fourth servomotors are configured to rotate the leg about the Z axis and the fifth and sixth servo motors rotate the leg about the Z axis and the X axis It is preferable to be constructed so as to be able.

The above and other objects and advantages of the present invention will become more apparent by describing in detail exemplary embodiments thereof with reference to the attached drawings in which: In addition to the principal task solutions as described above, various task solutions according to the present invention will be further illustrated and described.

The movement method according to the obstacle recognition using the camera and the infrared sensor according to the present invention comprises four legs having a six-joint structure and is configured to be able to move while being quickly attached to and detached from the wall, It is possible to realize quick and smooth movement and work regardless of the shape and condition of the vertical wall surface.

In addition, since the present invention is configured to recognize obstacles and to move along a wall while avoiding or overcoming obstacles by using a camera and an infrared sensor, it is possible to overcome the limitations of the conventional wall-mounted robot and to improve the usability and reliability of the wall- There is also an effect. That is, the present invention uses a sensor and a camera capable of measuring a distance to an obstacle in order to solve the limitation that a wall-moving robot can not overcome an obstacle, accurately recognizes the characteristics of the obstacle on the vertical wall surface, aligns the obstacle with the robot horizontally The obstacle can be stably crossed so that the robot can be easily moved even if it is not a flat wall surface, thereby improving the reliability and usability of the robot.

1 is a perspective view of a wall-moving robot over an obstacle using a multi-joint according to an embodiment of the present invention, as viewed from above.
FIG. 2 is a bottom perspective view of a wall-moving robot over an obstacle using a multi-joint according to an embodiment of the present invention.
FIGS. 3 to 6 are a plan view, a bottom view, a front view, and a side view of a wall moving robot over an obstacle using a multi-joint according to an embodiment of the present invention.
FIG. 7 is a detailed view illustrating a leg portion of a wall-mounted robot according to an embodiment of the present invention, illustrating six joints.
FIG. 8 is a detailed view illustrating a leg portion of a wall-mounted mobile robot according to an embodiment of the present invention, illustrating a suction mechanism.
9 is a circuit diagram illustrating a central control unit and a leg control unit of a wall-mounted robot according to an embodiment of the present invention.
10 is a circuit diagram showing a leg control unit of a wall-mounted robot according to an embodiment of the present invention.
11 is a reference diagram showing the recognition range of the ultrasonic sensor (a) and the infrared sensor (b).
12 is a view showing the results of inspection of the precision of an infrared sensor applied in the present invention.
13 is a reference view showing an example of an evasive obstacle in a vertical structure.
14 is a reference view showing an example of an obstacle to be overcome in a vertical structure.
15 is a view showing a result of image processing using a camera according to the present invention, and is a reference showing a type that is not detected as an obstacle.
16 is a view showing a result of image processing using a camera according to the present invention, and is a reference view showing various image processing results according to an obstacle detection type.
FIG. 17 is a view showing a result of image processing using a camera according to the present invention, and is a reference view showing an image processing result of a curved obstacle type. FIG.
FIG. 18 is a reference diagram for explaining a corresponding method according to the obstacle straight line judgment in the present invention.
FIG. 19 is a flowchart showing an obstacle recognition and corresponding method according to an embodiment of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings.

FIGS. 1 to 10 are views for explaining a wall-moving robot moving over an obstacle using a multi-joint according to an embodiment of the present invention. FIGS. 1 and 2 are a perspective view of the wall-moving robot viewed from above and from below, 6 is a plan view, a bottom view, a front view and a side view of the wall-moving robot, FIG. 7 is a view for explaining six joints of a wall-moving robot, FIG. 8 is an explanatory view of a suction mechanism of a wall- And a central control unit and a leg control unit of the robot.

The wall-moving robot according to the present invention has four legs 200 to minimize the overlap of the activity radii of the legs, and each leg 200 is configured to have six joints that can freely move on the vertical wall surface. Each of the legs 200 is configured to be capable of suction and movement by vacuum. It can be stably attached to a wall surface of a large structure having a smooth surface by using an absorption method by vacuum generated by rotating the vacuum motor 320 without an air compressor. In the attaching process, four infrared sensors 240 are used to accurately and safely attach the robot to the wall. Each leg 200 is configured to have a single central control unit 450 for controlling and increasing the speed of each leg control unit 410 to control the legs 200 and for the integrated control. The central control unit 450 is configured to integrally control each leg 200 so that the infrared sensor 240 and the camera 500 can recognize the obstacles present in the movement route and can avoid the obstacle.

The wall-moving robot over the obstacle using the multi-joint according to an embodiment of the present invention includes a body 100, four legs 200 connected to the body 100 and having a six-joint structure, A suction mechanism 300 for generating a vacuum force on the vacuum pads 310 provided at the ends of the four legs 200 to generate attraction force for attaching the legs 200 to the wall surface, And a central control unit 450 that integrally controls the legs of the legs 200. The control unit 400 includes a servomotor 210 provided in the legs 200 and a leg control unit 410 provided in each leg 200 to control the attracting mechanism 300, .

The main components of the mobile robot will be described in detail as follows.

Referring to FIGS. 1 and 2, the body 100 includes two plates 110 and 120 spaced apart from each other in the vertical direction, and four legs 110 and 120 are formed at corner portions of the two plates 110 and 120, (200) are respectively assembled. It is preferable that the corner portions where the four legs 200 are assembled in the two plates 110 and 120 are configured to have a relatively more protruding structure so that the movement of the legs 200 can be smoothly performed without interference.

A control device 400 may be provided on the upper part of the body 100 and a camera 500 and an infrared sensor 550 may be provided on the front side to recognize obstacles.

In the present invention, the body 100 is configured to be smaller than the size of the entire wall-mounted robot, for minimizing the moment, distributing the load, and reducing the volume. This is because when the body size increases, body sagging occurs due to load concentration, and as the weight increases, the moment of the body occurs, and the volume of the body when the obstacle moves on the wall infects the autonomy and activity of the robot .

The vacuum motor 320, the infrared sensor 240, the solenoid valve 340, the solenoid valve 340, the solenoid valve 340, the solenoid valve 340, And the like on the leg 200 side to disperse the load to minimize the moment and the volume that the body 100 can have. As a result, the sagging phenomenon of the body is reduced and interference with the movement of the robot can be reduced. As the weight of the body decreases, the moment of the body decreases, and the distortion of the body can be reduced. In addition, as the volume is reduced, the control of the shaft is free, the maximum flexibility, expandability, autonomy, etc. in the minimum space can be improved and the movement on the wall can be facilitated.

Next, the legs 200 are configured in a six-joint structure so as to overcome an obstacle as well as walking on a vertical wall surface. Further, the vacuum pad 310 to be described below has a six-joint structure in which the degree of freedom of the vacuum pad provided on each leg 200 can be increased because the contact surface does not adhere properly if the contact surface is slightly deviated.

The leg 200 includes six servomotors 210 M1 to M6 and respective brackets 220 for supporting the servomotors 210 and for coupling the axes of the servomotors 210 to each other. do. The axis of the No. 1 servo motor 210 connected to the body 100 is assembled directly between the two plates 110 and 120 constituting the body 100.

Referring to FIG. 7, the role of the six servo motors 210 will be described. The No. 1 servo motor M1 connected to the body 100 rotates the Y axis to move the body 100 of the robot back and forth And the second servomotor M2 serves to adjust the angle of the leg 200 and to attach the vacuum pad 310 by performing X-axis rotation for movement on the wall surface where the inclination changes . The servo motors M3 and M4 take charge of the rotation of the Z axis so as to lift the leg 200 of the robot so as to overcome obstacles and the servo motors M5 and M6 of the fifth and sixth servo motors M5 and M6, So that the vacuum pad 310 can be attached at an accurate angle by rotating the Z-axis and X-axis.

The six-joint-structure leg 200 imitates a spider that can perform a universal movement even in a special environment such as overcoming a slope, moving a wall, and obstacles on a wall, and is equipped with six servomotors 210), thereby overcoming the problem of the limited degree of freedom of the robot model.

The six servomotors 210 are preferably composed of motors capable of rotating up to 300 degrees and capable of controlling the rotation speed so that they can sufficiently perform their role as joints. In order to eliminate the daisy chain phenomenon, It is preferable to supply the voltage and current more stably using the battery so that the circuit is not damaged.

The bracket 222 on the side of the vacuum pad 310 among the brackets 220 constituting the joint is preferably formed to have a long arc shape when the leg 200 is viewed from the side. This is so that the vacuum pad 310 is easily accessible horizontally when it is attached to the wall surface.

In addition, it is preferable that the four legs 200 are equipped with an attachment distance measuring sensor for measuring the distance so that the vacuum pad 310 is attached while approaching the wall horizontally. The adhesion distance measuring sensor preferably includes four infrared sensors 240 installed around the vacuum pad 310.

Four infrared sensors 240 may be installed at four corner portions in a rectangular structure at the top of the vacuum pad 310 as illustrated in the figure. A sensor support 230 may be provided on both sides of the bracket 220 to which the vacuum pad 310 is connected and an infrared sensor 240 may be installed on both ends of the sensor support 230 .

In order to accurately attach the vacuum pad 310 of each leg 200 to the wall surface, the slope between the pad and the wall surface must be accurately known since the pad and the wall surface must be horizontal so that the vacuum pad 310 can be attracted most stably. Since four infrared sensors 240 are provided on the vacuum pad 310 of each leg 200 in the present invention, the distance between the wall surface and the vacuum pad 310 is set to be And if the measured value is larger than a predetermined error, it is determined that the motor is inclined, and accordingly, the servomotors M5 and M6 of No. 5 and No. 6 can be moved to adjust the level.

When the infrared sensor 240 is attached without using the infrared sensor 240, it is possible to accurately determine whether the infrared sensor 240 is attached to the pressure sensor 330 described below. However, if the infrared sensor 240 is not attached, Additional control should continue. However, when four infrared sensors 240 are used, it is possible to apply the algorithm that perfectly adheres the leg 200 from the beginning, so that not only the attachment speed can be increased but also the width of the obstacle width can be recognized even when crossing the obstacle The obstacle can be more easily passed over.

Next, the suction mechanism 300 is configured to suck air in the vacuum pad 310 by using the vacuum motor 320 to make the inside of the pad into a vacuum state, and to quickly and accurately attach and detach the leg 200.

8 and the like, the suction mechanism 300 includes a vacuum pad 310 provided at the end of the leg 200 and attached to a wall surface of the vacuum pad 310, a vacuum pad 310 connected to the vacuum pad 310 by a hose 350, A vacuum motor 320 for generating a vacuum on the vacuum pad 310 and a vacuum pad 310 provided on one of the vacuum pad 310, the hose 350 and the vacuum motor 320, A pressure sensor 330 and a solenoid valve 340 disposed on the hose 350 to remove vacuum pressure formed on the vacuum pad 310.

Here, a fitting 360 having a cruciform structure may be used to connect the hose 350 connected to the vacuum pad 310, the vacuum motor 320, the pressure sensor 330, and the solenoid valve 340. At this time, the fitting 360 may be installed on the bracket 220 or the upper portion of the servo motor 210, and it is preferable to use a two-touch fitting.

It is preferable that the vacuum pad 310 is formed in a flat and wide shape so that the vacuum pad 310 is in a vacuum state even if only a small amount of air is sucked and the suction force is increased.

The pressure sensor 330 functions to accurately determine whether the vacuum pad 310 is accurately attached to the wall surface by measuring the vacuum pressure. That is, when the measured value of the pressure sensor 330 is greater than a predetermined value, it means that the vacuum pad 310 is attached, and when the measured value is less than a predetermined value, the pad is not attached. Therefore, if the value of the pressure sensor 330 does not increase for a predetermined time or more when the leg 200 is lowered to the wall surface, it can be determined that the vacuum pad 310 is not attached. The algorithm may be a reattachment method or a method of accurately attaching the leg 200 using another sensor.

The solenoid valve 340 functions to release the vacuum pad 310 from the wall surface by opening the vacuum pad 310 with vacuum under the atmospheric pressure condition to quickly release the vacuum.

The pressure sensor 330 and the solenoid valve 340 are installed on the bracket 220 at the end of the leg 200. The pressure sensor 330 and the solenoid valve 340 are installed in the bracket 220 at the end of the leg 200, .

Accordingly, the adsorption mechanism 300 creates a space blocked from the outside by using a blocking film of the vacuum pad 310 for attachment to the wall surface, and rotates the vacuum motor 320 to exhaust the air inside the vacuum pad 310 The vacuum state is maintained. At the same time, the vacuum state is confirmed by the pressure sensor 330 to ensure the stability of the attachment. In order to move the leg 200, it is necessary to remove the vacuum, and the solenoid valve 340 adjusts the pressure between the atmospheric pressure and the vacuum pad 310 so as to move quickly and enable quick movement.

Next, the control mechanism has four leg control portions 410 for controlling the movement of each leg 200, and each leg control portion 410 is configured so as to be controllable by the central control portion 450.

The control mechanism 400 includes four leg control units 410 for controlling the servomotors 210 on the legs 200 and the adsorption mechanism 300 provided on the legs 200, And a central controller 450 for receiving signals from the four leg controllers 410 and integrally controlling the four leg controllers 410.

Referring to FIG. 10, each leg control unit 410 can be configured using Arduino Nano, which is small in size. The leg control unit 410 has two functions of attaching and detaching the wall for robot movement and transmitting information to the central control unit .

Each of the legs 200 includes four infrared sensors 240, one pressure sensor 330, one vacuum motor 320, one solenoid valve 340, and six servo motors 210, 13 sensors and motors, each of which is configured to receive measurement signals from 13 sensors and motors and to output control signals. That is, the moving position is controlled by the servo motor 210 using the information received from the central control unit 450 to move the robot. The wall surface and the vacuum pad 310 are horizontally aligned by using the four infrared sensors 240 while the vacuum motor 320 is ON and the solenoid valve 340 is OFF, To check whether or not the attachment is successful. The pressure in the vacuum pad 310 is checked using the pressure sensor 330 in the OFF state of the solenoid valve 340 and the vacuum motor 320 in the OFF state, Lift the legs (200).

Also, the leg control unit 410 plays a role of transmitting the information of the rotor of the central control unit 450 itself. That is, it is configured to send the values of thirteen sensors and motors of each leg 200, the vacuum state information of the vacuum pad 310, and the operation state information of the legs 200 to the central control unit 450.

The central control unit 450 controls the leg control unit 410 and the obstacles to be described in detail below, and performs image processing using openCV, bridge modeling using openGL, GUI implementation using QT, .

The central control unit 450 can be configured using the Jetson TK1 Linux board. By using Linux board, it is possible to solve the image processing problem that requires a lot of processing, and it is more versatile than other control parts. Using QT programming, it is possible to program Linux GUI, which allows users to display values or visualize them. By using OpenGL, motion of real servo motor can be displayed and displayed graphically, and image processing can be smoothly performed using OpenCV. You can also implement parallel processing more flexibly using threads or forks.

The core function of the central control unit 450 is configured to realize synchronization of operations through the four leg control units 410. [ That is, since the bridge control unit 410 takes charge of the start and end of communication, the central control unit 450 generates a thread that continuously waits for communication with the bridge control unit 410 and performs communication. And performs synchronization and control through the data received from each leg control unit 410.

 In addition, the central control unit 450 is configured to provide an obstacle recognition and coping function. That is, the obstacle is recognized by using the camera 500 and the infrared sensor 550 provided in front of the body 100. The presence or absence of an obstacle is determined through the infrared sensor 550. If the measured distance is within 30 cm, the image processing of the camera 500 can be applied. It is possible to recognize and cope with the obstacle by using a method of applying the image processing of the camera 500 to determine whether the obstacle is an obstacle, or to avoid the obstacle. These obstacle recognition and control methods are described in detail below.

Since the one leg control unit 410 controls the six servo motors 210 by using the central control unit 450 and the four leg control units 410 as described above, The reaction rate and the processing speed can be increased by about four times and the respective legs 200 can be operated simultaneously as the processing speed of the servo motor 210 of each leg 200 increases. In addition, the central control unit 450 can additionally implement functions that are impossible with the conventional MCU such as image processing, GUI, and TCP / IP communication by using a Linux board.

Now, an obstacle recognition method of the wall-mounted robot according to the present invention constructed as described above will be described in detail with reference to FIG. 19 and the like.

In order to accurately identify obstacles, it is necessary to consider the distances between the robot and the obstacles as well as the inclination between the obstacles and the sensors that are trying to measure them. In particular, when a multi-joint robot attached to a vertical wall crosses an obstacle, the robot and the obstacle must be horizontal so that the body can be stably stuck to the obstacle without crossing the wall.

Therefore, in the present invention, a method of recognizing an obstacle on a vertical wall surface using the infrared sensor 550 when the robot faces an obstacle, and a method of confirming whether or not an obstacle is avoided using the camera 500 will be described.

In the obstacle recognition, it is determined whether the obstacle is a curve or a straight line, and in the case of a straight line, the infrared sensor 550 described above is used to horizontally align the robot and the obstacle so that the obstacle can be stably crossed.

For reference, the most important thing in a distance measuring sensor is the orientation angle. Since the inclination between the obstacle and the sensor for measuring the obstacle is sensitive, the larger the direction angle, the wider the detection width, and the error of detecting the length value of the surrounding object rather than the obstacle is likely to occur. In order to prevent this, the orientation angle should be kept to a minimum. 11, the recognition range of the infrared sensor b shown in FIG. 11 (b) is narrower than the recognition range of the ultrasonic sensor shown in FIG. 11 (a).

The infrared sensor used for the obstacle recognition used in this embodiment can use GP2Y0A41SK0F.

 The infrared sensor 550 emits infrared rays from the light emitting portion and receives the infrared rays from the light receiving portion. When external infrared rays enter the light receiving part between them, noise as shown in FIG. 12 is generated and the accuracy of the measured distance can not be guaranteed every time. Therefore, in order to remove noise, n distance value is read, and the values are sorted, and a value having a difference of more than the threshold e from the median is judged as noise. e is the error tolerance specified by the user. An average of the values not judged as noise is obtained to obtain the final distance.

The algorithm for removing noise can be set as follows.

이라고 표현한다.(1) read data by the desired number n and collect it

.

는 집합

을 정렬한 값들의 집합이다.(오름차순 또는 내림차순)(2) set

Is a set

(Ascending or descending order).

에서 사용자가 설정한 오차 허용 범위 e의 ±값을 통해 값을 정제한다. 그것을 집합

라고 하면,

는 다음 [수학식 1]과 같다.(3) set

The value is refined by the ± value of the error tolerance range e set by the user. It is set

In other words,

Is expressed by the following equation (1).

[Equation 1]

에 속한 모든 원소들의 평균은

라고 정의한다.(4)

The average of all elements in

.

(5) The standard deviation s is expressed by the following equation (2).

&Quot; (2) "

(6) The 95% confidence interval is defined through the following formula (3) through the defined standard deviation s.

&Quot; (3) "

는 다음의 [수학식 4]과 같이 집합

의 모든 원소 중

함수에 참인 값들의 집합으로 정의한다.(7) set

Is expressed by the following equation (4)

Of all elements of

It is defined as a set of values that are true to the function.

&Quot; (4) "

는 집합

의 평균으로 정의한다. 이를 [수학식 5]로 나타내면 다음과 같다. (8) Distance value to be obtained

Is a set

. This can be expressed by Equation (5) as follows.

&Quot; (5) "

13 and 14 are photographs showing vertical wall obstacles. As shown in the figure, the black arrow indicates the direction in which the robot looks at the obstacle, the white arrow indicates the direction of passing over the obstacle, and L means the width of the obstacle.

In both cases shown in FIGS. 13 and 14, an obstacle is recognized through the infrared ray sensor 550. When the width of the obstacle is measured through the image processing of the camera 500, it can be seen that the width of FIG. 13 is narrow and the width of FIG. 14 is wide.

The robot of the present embodiment grasps the width of the obstacle and avoids the case where the width is narrower than the set value as shown in FIG. That is, the robot is moved in a way to avoid obstacles. However, if the width is wide as shown in Fig. 14, the robot is moved in such a way that it crosses obstacles.

When the image data is read through the camera 500, the image is converted into grayscale through Gray Level Transformation. The boundary line is detected by applying the Canny Edge Detection Algorithm to the image created in grayscale. Apply a leveling algorithm to the detected image to draw a line between the starting point and the ending point.

Obstacles can be largely divided into straight obstacles and curved obstacles. 15 and 16 show camera original images and boundary line extraction results according to various types of linear obstacles.

가 0이거나 w,

는 0이거나 h이어야 한다. 장애물을 검출 할 때, 시작점과 끝점은 한 개씩 있어야 하고 시작점과 끝점이 여러 개인 경우에는 회피 하는 장애물이라고 판단한다. The key to detecting whether or not an obstacle is a starting point and an end point. For the start and end points to be established, if w is the width of the image and h is the height of the image,

Is 0 or w,

Must be 0 or h. When detecting obstacles, there must be one starting point and one ending point, and it is judged to be an obstacle to avoid when there are a plurality of starting and ending points.

In the case of FIG. 15, since the starting point and the ending point are not established, it can not be judged to be an obstacle. Therefore, it is judged to be an obstacle to avoid.

16 shows a case where a start point and an end point are established. In the case of (a), (b), and (c), the starting point and the ending point are located at the left and right ends. (d), it is judged that the width is narrow and there should be avoided because the starting point and the ending point exist at the top and bottom.

이라고 하고 끝점을

라고 했을 때,

을 이용하여 기울기를 알 수 있다. 도 16의 (a)는 기울기가 0이므로 회전하지 않는다. 그러나 도 16의 (b)와 (c)의 경우는 기울기가 0이 아니므로 회전한다. 도 16의 (b)와 (c)를 보면 흰색 선이 일정 하지 않고 좁아지는 구간과 넓어지는 구간이 있다. 회전은 구간이 좁아지는 곳에서 넓은 곳으로 해야 한다. (b)의 경우는 오른쪽으로 회전하고 (c)의 경우는 왼쪽으로 회전해야 한다.In case of an obstacle to be crossed, the robot and the obstacle must be level to be able to safely cross the obstacle. Starting point

And end point

When you say,

Can be used to know the slope. 16 (a) does not rotate because the slope is zero. However, in the case of FIGS. 16 (b) and 16 (c), the slope is not 0, and thus the rotation is performed. 16 (b) and 16 (c), there is a section where the white line is not constant but narrowing section and a section where the white line is broadened. Rotation should be from wide to narrow. (b) should be rotated to the right and (c) should be rotated to the left.

을 구한다.

과 사용자가 정한 e(오차 허용 범위)값을 비교하여 비교 값이 크다면 곡선으로 판단한다. e의 값이 낮아질수록 정확도는 높으나 직선을 곡선으로 분류할 수 있어 하드웨어에 따른 실험이 필요하게 된다. 다음 [수학식 6]은 곡선 판단 수식을 나타낸다. Fig. 17 is a curved obstacle and the condition of the start point and the end point is established. Select the start point and the end point and the line is A. To determine whether a curve is a curve, we divide the value of x by a constant value n from 0 to w to capture the baseline. FIG. 17 shows an example in which a reference line is drawn when n is 10, and the line is drawn as a line. By comparing the curve A with the straight line A on the 10 baselines

.

And e (error tolerance) values determined by the user are compared. If the comparison value is large, it is determined as a curve. The lower the value of e, the higher the accuracy, but the straight line can be categorized as a curve, which necessitates an experiment depending on the hardware. The following equation (6) represents a curve judgment formula.

&Quot; (6) "

As described above, since the image processing in the present invention generates a large load in the embedded system, if the infrared sensor 550 determines that there is an obstacle, the image processing starts.

를 알 수 있고 로봇의 회전각도

는 모든 로봇이 알고 있다. 거리 값

와 회전각도

를 알게 되면 삼각함수를 이용하여 회전 후 거리 값

를 예상할 수 있다.

를 통해 나온 예상 값

와 실제 회전 후 측정된 거리 값

를 비교하여 직선인지 다시 한 번 판단할 수 있다. 다시 한 번 판단하여 나온 결과가 직선이라면 문제가 없지만 다를 경우에는 곡선임을 판단하고 회피할 수 있도록 한다.First, the type of the obstacle is determined using the camera 500. The judgment using the camera depends on e (error tolerance), so it can be judged as a straight line. Therefore, there is a need for a method to check whether the line is straight or curved more precisely. Figure 18 shows this method. The place you are looking at before the rotation is represented by a dotted line, and the place where you are looking after the rotation is represented by a solid line. Before the rotation,

And the rotation angle of the robot

All robots know. Distance value

And rotation angle

If you know the distance,

Can be expected.

Estimated through

And the measured distance after actual rotation

And it is possible to judge again whether or not the line is straight. If the result of judging again is a straight line, there is no problem, but if it is different, it is judged to be a curve and it can be avoided.

This method shows an obstacle recognizing method optimized for a robot moving on a vertical wall surface by using the camera 500 and the infrared sensor 550. The camera 500 and the infrared sensor 550 to determine the distance to the obstacle, the inclination of the obstacle, the possibility of avoidance, and the presence or absence of the curvature of the obstacle.

In the obstacle recognizing method of the wall-mounted mobile robot according to the present invention, the distance measuring sensor uses an infrared sensor 550 having a narrow steering angle, and uses an algorithm to secure the accuracy of the infrared It is increasing. In addition, the camera 500 and the control unit can be used as a high-precision sensor to control the movement of the robot by detecting the characteristics of various obstacles by boundary line extraction and curve / straight line judgment on the image.

As described above, the technical ideas described in the embodiments of the present invention can be performed independently of each other, and can be implemented in combination with each other. While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is evident that many alternatives, modifications, and variations will be apparent to those skilled in the art. It is possible. Accordingly, the technical scope of the present invention should be determined by the appended claims.

100: body 110, 120: plate
200: Bridge 210: Servo motor
220: bracket 230: sensor bracket
240: Infrared sensor 300: Absorption mechanism
310: Vacuum pad 320: Vacuum motor
330: Pressure sensor 340: Solenoid valve
400: control mechanism 410: leg control unit
450: central control unit 500: camera
550: Infrared sensor

Claims (6)

A first step of determining whether there is an obstacle ahead by using an infrared sensor provided in the robot and calculating a distance to the obstacle in the presence of the obstacle;
If it is determined that there is an obstacle in the first step, the width of the obstacle is measured through image processing of the obstacle taken by the camera, thereby avoiding the case where the width of the obstacle is narrower than the set value. If the width of the obstacle is larger than the set value, And a second step of moving the robot in a passing manner,
The method of calculating the distance in the first step includes: reading the distance value of n times, sorting the values, removing a value having a difference equal to or larger than the tolerance range e specified by the user from the median value as noise, A method of obtaining the final distance by obtaining an average of the values that have not been obtained,
(1) read data by the desired number n and collect it

And,
(2) set

Lt; RTI ID =

To a set of sorted values,
(3) set

, The value is refined through the ± value of the error tolerance range e set by the user,

Lt; / RTI >
(4)

The average of all elements in

By definition,
(5) The standard deviation s is calculated by the following equation (2)
(6) The 95% confidence interval is calculated through the following equation (3) through the standard deviation s,
(7) set

(4) using the following equation

Of all elements of

Define the function as a set of values that are true,
(8) Distance value to be obtained

Is a set

And calculating the distance through the following formula (5): " (5) "
[Equation 1]

&Quot; (2) "

&Quot; (3) "

&Quot; (4) "

&Quot; (5) "

delete delete The method according to claim 1,
And moving the robot by a slope when the slope of the obstacle is not 0 in the second step. The method according to claim 1,
The robot includes a body; Four legs each connected to the body and having a six-joint motion structure by respective motors; An adsorption mechanism for forming vacuum pressure on a vacuum pad provided at an end of each of the legs to generate an attraction force for attaching each leg to a wall surface; And a controller for controlling the motors and the suction mechanism of each of the legs. The moving robot according to claim 1, The method of claim 5,
The legs include six servomotors that are capable of six-joint operation,
The first servo motor is configured to rotate the leg about the Y axis with respect to the body, the second servo motor is configured to rotate the leg about the X axis, and the third and fourth servo motors And the 5th and 6th servo motors are configured to be able to rotate the legs about the Z axis and the X axis. The obstacle of the robot for moving the wall using the camera and the infrared sensor Moving method according to recognition.

Images