KR20100117931A

KR20100117931A - Method of generating sweeping work path for mobile robot

Info

Publication number: KR20100117931A
Application number: KR1020090036659A
Authority: KR
Inventors: 현웅근
Original assignee: 호남대학교 산학협력단
Priority date: 2009-04-27
Filing date: 2009-04-27
Publication date: 2010-11-04

Abstract

The present invention relates to a method for generating a sweeping work path of a mobile robot, the method comprising: selecting a work area for a mobile robot to work on a drawing created and loaded by an autocad; and an obstacle recognized in a shape space for the selected work area. Generating an obstacle map for the target; generating an intermediate target point for generating a primary sweeping path for the work area using obstacle information; generating a primary sweeping path using the intermediate target points; Generating a collision avoidance path for avoiding collision with an obstacle when the collision with an obstacle occurs on the primary sweeping path using a map and mobile robot information, and first sweeping the generated collision avoidance path And generating a final sweeping work path for the mobile robot by reflecting on the path. According to the sweeping path creation method of the mobile robot, a work area can be set using the drawings created by AutoCAD as it is, and a sweeping work path capable of performing work while avoiding collision with an obstacle can be easily calculated have.

Description

Method of generating sweeping work path for mobile robot}

The present invention relates to a method for generating a sweeping work path of a mobile robot, and more particularly, to a method for generating a sweeping work path of a mobile robot, which enables the mobile robot to generate a sweeping path to work by using a drawing made in CAD.

In recent years, the application of robots has been expanded from the use in the factory to the general work. In particular, there is a strong demand for robotization of work such as normal inspection of nuclear facilities, plastering for construction work, cleaning work, and painting. The path plan for such a task requires a motion plan that moves around the empty space of the 2D surface area, which is a work area, and is called a sweeping task path plan. The action planning method for the sweeping work is largely classified into an on-line method and an off-line method. The online method is a method in which the robot moves around the work area and recognizes the work environment as a sensor and reflects it directly in the motion. However, since this online work method does not use the entire map during the work, the robot cannot guarantee the completion of the sweeping work within a given time, and the work ability depends on the shape of the environment. In addition, depending on the obstacle environment, there is a problem of falling into a dead-lock phenomenon, which hoveres around the same area. Due to these problems, most existing robots that do actual plastering have only a function of repeating a certain movement pattern and are semi-automatic type controlled by a functioning operator. The offline method is a method of creating a swept work path on the assumption that you know information about the environment in which you will work. A method for automatic generation of a sweeping work path is proposed by D. Kurabashi et al., Which is a method for creating a sweeping work path for a cleaning robot and a plaster robot, This is how the diagram is applied. In this method, the Voronoi diagram generates the non-collision path in the configuration space, but in the Cartesian coordinate system, it is created by connecting the center line or the point between the obstacle and the workspace, so that the generated path is too close to the obstacle The problem may occur that the path control needs to be elaborated while the actual robot is running, and in some cases, the actual robot cannot go the collision-free path generated between the obstacles.

The present invention was devised to improve the above problems, and an object thereof is to provide a sweeping work path generation method of a mobile robot that can easily generate a sweeping path while avoiding obstacles from the work environment map drawing information.

Still another object of the present invention is to provide a sweeping work path generation method of a mobile robot, which can easily generate a sweeping work path using a drawing made in an autocad.

In order to achieve the above object, a sweeping work path generation method of a mobile robot according to the present invention is provided. Selecting a work area for the mobile robot to work on a drawing created and loaded by autocad; I. Generating an obstacle map for the obstacle recognized in the shape space for the selected work area; All. Generating an intermediate target point for generating a primary sweeping path for the work area using the obstacle information; la. Generating the primary sweeping path using the intermediate target points; hemp. Generating a collision avoidance path for collision avoidance with an obstacle when a collision with an obstacle occurs on the primary sweeping path using the obstacle map and the mobile robot information; bar. And generating a final sweeping work path for the mobile robot by reflecting the collision avoidance path generated in the step D in the first sweeping path.

Wherein the intermediate target points are positional information that the mobile robot scans in a zigzag form in the work space at a predetermined width and makes contact with a wall or an obstacle when the mobile robot moves while the obstacle information is in contact with the scan direction of the mobile robot Direction, that is, whether it is left or right, and obstacle attribute information, that is, whether it is a wall or an intermediate obstacle.

According to the method for generating a sweeping work path of a mobile robot according to the present invention, a work area can be set using the drawings created by the CAD as it is, and the sweeping work path that can perform the work while avoiding collision with obstacles can be easily performed. Can be calculated.

Hereinafter, a method of generating a sweeping work path of a mobile robot according to a preferred embodiment of the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram illustrating a sweeping work path generator of a mobile robot according to the present invention.

Referring to FIG. 1, the sweeping work path generator 10 of the mobile robot includes a CAD interface module 20, an obstacle mapping module 30, an intermediate target point generation module 40, and a sweeping work path generation module 50. An avoidance path generation module 60, an input unit 61, a display unit 65, a main control module 70, a mobile robot drive module 80, and a storage unit 90 are provided.

The CAD interface module 20 provides a menu for selecting a work area setting and a robot type from a drawing of a CAD file created and loaded with AutoCAD and a link with AUTO CAD. Data corresponding to the region is extracted to be used by the sweeping work path generator 10.

The obstacle mapping module 30 generates an obstacle map from obstacles recognized from the selected work area data.

The intermediate target point generation module 40 generates the intermediate target point SP required for generating the sweeping path.

The sweeping work path generation module 50 generates the first sweeping work path using the intermediate goal point information generated by the intermediate goal point generation module 40, and then the collision avoidance path generated by the collision avoidance path generation module described later. The information is used to generate the final sweep path.

The collision avoidance path generation module 60 generates a collision avoidance path in the work area by using the mobile robot 100 and obstacle information to be applied.

The input unit 61 supports selection of a work area, a robot model, and the like.

The display unit 65 is controlled by the main control module 70 to display display information.

The main control module 70 controls each module 20 to 60 to generate a sweeping work path.

The storage unit 90 stores data generated in the process of sweeping work path generation.

The mobile robot drive module 80 generates a driving command for driving the mobile robot 100 to correspond to the last generated sweeping work path.

Hereinafter, a sweeping work path generating process will be described with reference to FIG.

First, the work area for the mobile robot to work from the drawings created and loaded by the AutoCAD is selected using the input unit 61 with the aid of the CAD interface module 20 (step 210).

Next, the obstacle is recognized on the drawing of the selected work area, and the obstacle map is generated in the shape space with respect to the recognized obstacle (step 220).

Thereafter, an intermediate target point for generating a primary sweeping path for the work area is generated from the obstacle map (step 230), and a primary sweeping path is generated using the intermediate target points (step 240).

When collision with an obstacle occurs in the process of generating the first sweeping path, a collision avoidance path is generated using the obstacle map and the mobile robot information to avoid the collision with the obstacle, and the generated collision avoidance path is reflected in the primary sweeping path. Thereby creating a final sweeping work path for the mobile robot (step 250).

Finally, the mobile robot 100 driving command is generated from the generated final sweeping path (step 260).

Here, the intermediate target points are position information of contact with a wall or an obstacle when the mobile robot 100 moves while scanning the moving robot in a zigzag form within a set width, and the obstacle information is a direction corresponding to the scanning direction of the mobile robot. Contains obstacle attribute information.

Hereinafter, this sweeping work path generation process will be described in more detail.

First, since the sweeping operation is performed on a two-dimensional plane, the mobile robot 100 and the target work are modeled on the two-dimensional plane. The mobile robot 100 is modeled as an arbitrary polygon, and omni-directional motion and self-rotating are applied. The shape of the mobile robot 100 supports the user to select or model.

In addition, the objects in the work space are modeled in the form of polygons.

Obstacles in the working environment are divided into polygonal shapes and curved shapes. The polygonal shape has information about each vertex, and the curved obstacle is divided into straight line segments having a constant resolution and displays a curve as a set of points. When displayed on the screen, curve fitting is performed to display curved obstacles. The data structure for the obstacle is as follows.

Struct object {

double * x_position;

double * y_position;

double radius_x, radius_y;

int edge_No;

char status;

} * object_posture;

Here, status is a status flag indicating that the obstacle is a polygon, a circle, an ellipse or a curve, and edge_No represents the number of vertices when the object is a polygon. In the case of a circle or an ellipse, variables of radius_x and radius_y are used. In case of an arbitrary curve, the shape and position data are equally divided into a constant segment having sufficient resolution.

The area to be worked on is the two-dimensional Euclidian space

To be displayed.

To

In the j- th stationary stationary object, that is, a fixed obstacle, an obstacle region in a configuration space is represented by Equation 1 below.

Where C is ----- ,

Is the position x, y and posture of the mobile robot

Is a set representing

Is the position and attitude of the mobile robot 100 determined by x, y, and q .

Given the number of fixed obstacles in p , the total space in the obstacle area is

Free space without obstacles in shape space

Is expressed by Equation 2 below.

=

The symbol "＼" here means subtraction of the set.

For all q included in

A region consisting of

If this is called a sweeping work area,

Silver sphere

Within the robot's physically accessible area, the sweeping

Performed in space. Under that condition,

Is satisfied, the sweeping operation is satisfied. Where d (τ, A (q)) is the path

And robot Indicates the distance between.

Where r is the scanning width when the robot sweeps.

Next is a mobile robot 100

Create a continuous path that sweeps.

First, the AUTO CAD interface module 20 provides an interactive menu using the CAD language Auto LISP and the DCL (Dialog Control Library) method, and extracts the robot type and selected work area data.

The AUTO CAD interface module 20 obtains the actual data to be used by the sweeping work robot through the drawing information retrieved using the Entity of the CAD and outputs the information data of the obstacle (line, circle, polygon, arc, etc.) .

The Obstacle Map Building module 30 is used for x-, y-, and 2-dimensional robots to be applied using obstacle and work environment data obtained from the CAD interface module 20.

3D Free Configuration Space

To form.

Subgoal Generation module 40

Generates an intermediate target point according to the collision form of the obstacle and the mobile robot (100).

The collision avoidance path generation module 60 generates the obstacle collision avoidance path with respect to the collision path generated in the process of generating the primary sweeping work path by the sweeping work path generation module 50. That is, from the first sweeping work path described above

Creates a collision avoidance path at the shortest distance to the target point while avoiding obstacles.

The intermediate target point generating module 40 scans the work area with a scanning width r for the work space on the two-dimensional plane, and stores the contact points between the robot and the obstacle during the scan and the contact points between the robot and the work area. The smaller the scan width r , the smaller the sweeping width and the more overlapping areas, but more closely cleaning or plastering.

While the robot scans the selected work area, it saves the location of the robot, its obstacles, its contact with the wall of the work area and its collision state. This position is used as an intermediate target point for the sweeping work path. The data structure of the intermediate target point for storage is as follows.

Struct Subgoal {

int position [DIMENSION];

char status;

char tracked_flag;

int obstacleNo;

} * SubgoalPositions;

Where position [DIMENSION] is the contact position between the mobile robot 100 and the obstacle or the work area wall, and the status is the contact type. In this case, it is expressed as RIGHT_WALL, LEFT_OBSTACLE when the left side of the robot contacts an obstacle, and RIGHT_OBSTACLE when it contacts the right side. And tracked_flag is displayed as TRACKED when the intermediate target point is selected and inserted into the list S to generate the sweeping work path, and UNTRACKED when not selected. And obstacleNo expresses the number of obstacles that the robot touched when generating the intermediate target point. The robot scans the working area from the top left to the bottom right to create the intermediate target point.

Keep it.

The process for creating intermediate target points is as follows.

SP is called the subgoal point and the i th intermediate point data list entry

Let's do it. Let S be a set of SPs, S = {S1, S2, ..., Si_max}. Where i_max is the maximum number of SPs stored in Si. The initial position of the robot is ( x, y ,

) = (0,0,0)

x,

If y is referred to as the X-axis and Y-axis scan width, respectively, the following steps 1A through 4A will be described with reference to FIG. 3.

Step 1A ( Step 1A) : Set x, y and i to 0, respectively. Then, the robot is moved to make contact with the left upper end in the work area, and the position is set as the initial position, and the x, y position

The status is indicated as LEFT_WALL (LW) and the tracked_flag as UNTRACKED. Then i is increased by one.

Step 2A ( Step 2A) : Move the robot along the x axis with the y-axis fixed. If the robot encounters an obstacle in the work area,

And the contact status is RIGHT_OBSTACLE (RO) and increases i. If the robot continues to move along the x-axis and touches the obstacle wall, record the contact state as RIGHT_WALL (RW). tracked_flag is recorded as UNTRACKED.

Step-3A (Step 3A): ten thousand and one robot is moved to the top and bottom of the step (Step 5A) will be described below meet the lower Y-axis in the workspace. Otherwise we'll move the robot along the Y axis

y. Then move the robot to the left along the X axis with the Y axis fixed. If the robot touches an obstacle in the work area, its position

Record the contact state as LEFT_OBSTACLE, and record the obstacle number at that time. Then, the robot is continuously moved to the wall of the right working area, and when the robot comes into contact with the wall, the contact position is recorded and the contact state is recorded as LEFT_WALL (LW). At this time, the Tracked_flag is recorded as UNTRACKED.

Step 4A (Step 4A): along the axis of the robot Y

Go to y and go to Step 1A.

Step 5A (Step 5A): writes the i to i-max is the maximum number of intermediate target point for the sweep work path.

4 shows an example of a work area including five obstacles, and FIG. 5 shows a result of finding an SP by applying the proposed intermediate target point generation process.

As a next step, when the process of generating the intermediate target point is completed, the generated intermediate target point is scanned to generate the first sweeping work path. In this case, for the intermediate target point generated in the intermediate target point generation process, first, all the intermediate target points are not connected at least twice, and second, the first sweeping work path formed by connecting the intermediate target points is shortened do. This process finds the next intermediate target point to be connected according to the contact state of the current intermediate target point. The process of finding the sweeping work path is performed by the process of CASE I to be described later when the state of the intermediate target point is LEFT_WALL or LEFT_OBSTACLE, and when the state of the intermediate target point is RIGHT_ WALL or RIGHT_ OBSTACLE.

How to create it is determined.

The list is scanned because the robot scans from the upper left to the lower right of the work area to create an intermediate target point.

The position data of the robot stored in the initial entry of the rear side raises the upper left of the work area, and the contact state at that time becomes LEFT_WALL. The process for creating the sweeping work path is as follows.

First, y_max is the workspace

Is the maximum area of the Y axis, x_max

Is called the list for the sweeping work path.

CASE 1 : If the contact status of the current intermediate target point is LEFT_WALL or LEFT_OBSTACLE, perform the following steps.

Step 1B :

Find an intermediate target point in space S whose contact status is RIGHT_WALL or RIGHT_OBSTACLE among SPs with the same y value in list S.

This case is the case shown in (A) in Fig. That is, since the current contact state of the robot is LEFT_WALL, the SP is searched in the list S and the SP with the contact state RIGHT_WALL on the same Y-axis line is searched to form a sweep path segment. As in the case of (B) of FIG. 3, if the contact state of the current SP is LEFT_WALL, the SP whose contact state is RIGHT_WALL or RIGHT_OBSRACLE is searched among the SPs having the same y position value in the list S.

Step 2B : Find and list the shortest SP among the SPs found in Step 1B

Insert it in If the found SP is marked as TRACKED and the robot moves from the current SP to the next SP to be connected, if the collision with an obstacle occurs, go to Step 6B which will be described later. Otherwise, go to Step 3B .

For reference, (B) in FIG. 3 is a case where the contact state of the current SP is LEFT_WALL and the SP of the next contact state has two SPs, which are RIGHT_OBSTACLE or RIGHT_WALL. In this case, the SP is selected as the SP to be connected to the SP whose distance is short and the generated path segment does not collide with the obstacle,

Insert it in

Step 3B : y ← y +

y, if y < 0, go to Step 5B .

Step 4B : Find the SP with contact state RIGHT_WALL or RIGHT_OBSTACLE on the same Y axis line and list the SP if the path segment has no collision.

. 3 (C), where the current contact state of SP is RIGHT_WALL and the y value is the scanning width

Increase by y to find the same contact.

Step 5B : Find the SP indicated by UNTRACKED in the S data list, and determine the SP that is closest to the current SP as the next connected SP. If a path from the current SP to the next SP to be collided with an obstacle is called a collision avoidance path generation module 60 to be described later. That is, as the path of FIG. 3D, the collision avoidance path generation module 60 is called in this case. FIG. 6 illustrates a case where the first sweeping work path is found without applying the collision avoidance path generation process to the obstacle condition of FIG. 4. In FIG. 6, five collision path segments are shown.

Step 6B : End the program if the number of connected SPs is equal to i_max . Otherwise the SP path list

And insert the corresponding tracked_flag as TRACKED. Depending on the status of the linked SP , go to Step 1B of CASE I or CASE II.

CASE II : When the contact status of the currently connected SP is RIGHT_WALL or RIGHT_OBSTACLE;

In the case of CASE II, RIGHT_WALL and RIGHT_OBSTACLE are replaced with LEFT_WALL and LEFT_OBSTACLE in Step 1B to Step 4B of CASE I, and the rest of the procedure is the same as Case I.

Next, a collision avoidance path generation process will be described with reference to FIGS. 7 and 8. FIG.

The collision avoidance path generation process uses a two-stage path planning method. This method is a hierarchical local search algorithm, which is the position of robot, x, y and the attitude of robot.

The path is planned in a grid-based configuration space in which a three-dimensional shape space represented by is divided into a grid of a constant size. A quantized cell is referred to as an obstacle cell when included in an obstacle and a free cell when not included in an obstacle. The path planning method first consists of a sum of several grids of the basic size (the minimum of three cells in a gridd three-dimensional shape space).

3

3 or 2

2

2 size cells) as one large cell, connecting these large cells to make a connection path between the large cell including the starting posture of the robot and the large cells including the target posture, Large cells that minimize the evaluation function expressed by distance from the target cell, distance from the cell of the starting posture, and the number of obstacle cells in the large cell are selected and connected. The path generated by the connected large cells is searched (step 310) and if there is a tangled cell, it is removed (step 320). Here, the rear entangled cell refers to a state in which the large cell connecting branch is divided into several branches as shown in FIG. 7 in constructing the passage with a large cell. In FIG. 7, the

cells

2, 3, and 4 correspond to the rear entangled cells. That is, in the list connecting cells, only one cell is connected to the head and the tail, but two or more cells are connected. This data structure is reconstructed into a set of cells that are not entangled by backtracking with the cell where the entanglement started as an intermediate target point.

In step 330, it is determined whether there is an entangled cell, and if it is not found, the path is searched for in the passage to the base cell (step 340). If it is determined in step 350 that the path search is not successful, the large cell that failed to generate the path is stored as a virtual obstacle cell (step 360), and the process returns to step 310. In contrast, if it is determined in step 350 that the path search is successful, a collision avoidance path is generated (step 370).

In other words, if there is an entanglement during the connection of a large cell, a entanglement without a entanglement is made by a retrace method, and then a final path is made of the connection of cells of a basic unit in a connection passage formed of the connection of a large cell. In addition, when the final path is not made in step 350, the large cells are stored as virtual obstacle cells, and from step 320, the cells are excluded from the connection of the large cells.

The collision avoidance path generated by the collision avoidance path generation process is illustrated in FIG. 9.

Meanwhile, in FIG. 6, in which the generation result of the SP of FIG. 4 is shown, collisions are shown in five places. The collision avoidance path generation process is performed in Step 5B of CASE I and CASE II. The result is shown in FIG. 9, and the result of the final sweeping work path considering collision avoidance is as shown in FIG.

In FIG. 10, the movement path represented by white is a plaster (or cleaning) work path, and the path represented by blue represents a path that the robot moves in the overlapping area of the plaster work (or cleaning work).

- Simulation

Workspace

Size is up to 21 m

The robot is 21m and modeled as a regular octagonal shape with one side 41.42cm. The shape space for the collision avoidance path is gridized by dividing the work area by 10cm. The number of basic cells in the shape space is 210

210 = 44,100. The shunt width was 75 cm.

Experiment 1

240 cm in working space

1, 2, 4, 6, and 9, as shown in FIGS. 11 to 15, by arbitrarily installing an obstacle of a square shape of 240 cm in height, The computation time for evasion path generation was analyzed. The final result of the path is as shown in Figs. 11 to 15, and the calculation time and analysis table are shown in Table 1 and Figs. 16 and 17 below.

Item
Of
Number of obstacles SP
produce
Count crash
Route
Count SP
produce
time Space
Creation time Creation of collision avoidance path
time Final path
Creation time One 70 One 3.516 6.593 1.593 11.703 2 86 5 5.384 8.131 2.362 15.879 4 106 5 8.571 11.099 4.560 24.231 6 122 6 11.538 13.736 11.923 37.198 9 160 10 14.725 16.978 13.846 45.549

As shown in Table 1 and FIGS. 16 and 17, the number of collision paths is not proportional to the number of obstacles. This is because the collision path is determined by the position of the obstacle. However, when the number of obstacles in the workspace is large, it shows that a lot of collision paths occur.

- Experiment 2

By applying the method proposed in the present invention, the generation of the sweeping path and the final path length were calculated. The working area is 4m in size

4 m and the size of the work robot is 30 cm.

18 to 20 are to create a sweeping path by applying the method proposed in the present invention. In addition, Table 2 below shows the generated path length.

In the case of Fig. 18 In the case of Fig. 19 In the case of Fig. 20 This method 3484.09 cm 2900 cm 2627.17 cm

The sweeping work path generated by the method proposed in the present invention is a direction parallel type moving in a zigzag shape as shown in FIGS. 18 to 20.

Experiment 3

The proposed sweeping work path generation method is applied to the actual construction drawings. The construction drawing was made with AUTO CAD R 14, and the interface menu with CAD SW was made with AUTO LISP. In the interface created with AUTO LISP, it is constructed to determine the shape of the robot and the size of the work area. Test results using the constructed work path generator are shown in FIGS. 21 to 23, and even when applied to the actual drawings, it can be seen that the sweeping work path is precisely generated by the proposed method.

1 is a block diagram illustrating a sweeping work path generator of a mobile robot according to the present invention;

2 is a flow diagram illustrating a process of generating a sweeping work path of a mobile robot according to the present invention;

3 is a diagram for describing a process of generating an intermediate target point for generating a sweeping path;

4 is a view showing an example of a workspace having five stores,

5 is a view showing an intermediate target point generated according to the present invention for the workspace of FIG.

FIG. 6 is a view illustrating a sweeping path generated without applying collision avoidance to the workspace shown in FIG. 5;

7 is a diagram schematically illustrating a connection process of an entangled cell;

8 is a flowchart illustrating a process of generating a collision avoidance path;

FIG. 9 is a view illustrating a collision avoidance path generated through a collision avoidance path generation process with respect to FIG. 6;

FIG. 10 is a view illustrating a sweeping path finally generated by applying the collision avoidance path of FIG. 9;

FIGS. 11 to 16 are diagrams showing a final sweeping work path generated by applying the method of the present invention when there are 1, 2, 4, 6, and 9 obstacles respectively,

FIG. 17 is a graph illustrating the number of intermediate target points generated and the number of collision paths for the obstacle condition of FIGS. 11 to 16.

FIG. 18 is a graph showing the calculation time for each item of the buried conditions of FIGS. 11 to 16.

19 to 21 are views showing the sweeping work path as a result of applying the method of the present invention to another obstacle condition,

22 is a view showing an example of a construction CAD drawing,

FIG. 23 is a diagram illustrating an example of a menu screen provided by an interface module interfaced with the construction CAD of FIG. 22.

FIG. 24 is a view illustrating a sweeping work path generated for the work area selected for the CAD drawing of FIG. 22.

Claims

end. Selecting a work area for the mobile robot to work on a drawing created and loaded by autocad;

I. Generating an obstacle map for the obstacle recognized in the shape space for the selected work area;

All. Generating an intermediate target point for generating a primary sweeping path for the work area using the obstacle information;

la. Generating the primary sweeping path using the intermediate target points;

hemp. Generating a collision avoidance path for collision avoidance with an obstacle when a collision with an obstacle occurs on the primary sweeping path using the obstacle map and the mobile robot information;

bar. And generating a final sweeping work path for the mobile robot by reflecting the collision avoidance path generated in the step d) in the first sweeping path.

The method of claim 1, wherein the intermediate target points are position information of contact with a wall or an obstacle when the mobile robot moves while scanning the mobile robot in a set width in a zigzag form, and the obstacle information is a scan of the mobile robot. Sweeping working path generation method of a mobile robot, characterized in that it comprises a direction and obstacle attribute information corresponding to the direction.