JPH10260724A - Map generating method for passage environment - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method with which an error can be cleared when the level of this error is comparatively low and line segment extracting method to the result of shape measurement is facilitated as well even when the errors of estimated values in the position and direction of robot due to a dead reconning method occur. SOLUTION: From the shape measured result 102 of passage environment letting a wall existent in position relation made orthogonal by a scanning type laser sensor loaded on a mobile robot and an estimated value 109 in the position and direction of this mobile robot provided by the dead reconning method, the line segment corresponding to the mutually orthogonal wall almost expressing the shape of this passage environment is extracted so that the map of passage environment required for this mobile robot can be generated.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for generating a map of a passage environment by a mobile robot equipped with a scanning laser sensor.

[0002]

2. Description of the Related Art In recent years, service robots that perform various types of transport work while autonomously traveling in an indoor passage environment such as a hospital have become widespread. The autonomous traveling means that the mobile robot travels without requiring guidance by a guide tape or the like. However, in order for the service robot to perform autonomous traveling, it is necessary to prepare a map of a passage environment in which work is to be performed and to teach the service robot to the service robot. here,
The map is information expressing walls and other obstacles existing in the passage environment by line segments. Conventionally, this map is created using a drawing tool (hereinafter, referred to as CAD) activated by a person on a computer in accordance with information such as architectural drawings. Further, as means for solving this problem, a mobile robot equipped with an appropriate sensor system is guided by a joystick or the like to a passage environment required for map generation, and the shape of the passage environment is formed by the sensor system. There has been proposed a method of generating a map of a passage environment required by the service robot using the data obtained as a result of the measurement and the technique of dead reckoning (Japanese Patent Laid-Open No. 7-110709).

[0003]

However, Japanese Patent Application Laid-Open No.
In the method of 110709, since the estimated values of the position and orientation of the mobile robot obtained mainly by the dead reckoning method are used, errors in the estimated values of the position and orientation that occur successively when the mobile robot is guided are caused by errors in the generated map. There is a problem that it greatly affects the accuracy of the data. Therefore, the shape measurement result of the passage environment obtained by the method of JP-A-7-110709 is different from the actual shape of the passage environment. In particular, on a two-dimensional map expressed in an XY rectangular coordinate system, the shape of a straight wall, which should be a straight line, becomes a curve in the actual measurement result, and the line segment extraction processing required for map generation is performed. It will be difficult. Therefore, the present invention can clear the error of the estimated value of the position and orientation of the robot by the dead reckoning method if the degree of the error is relatively small, and can extract the line segment from the shape measurement result. An object is to provide a method that facilitates processing.

[0004]

SUMMARY OF THE INVENTION According to the present invention, an operator guides a mobile robot equipped with a scanning type laser sensor to an indoor passage environment required for map generation by using a joystick or the like while using the laser sensor. The shape of the indoor passage environment is measured, and a map of the indoor passage environment required by the service robot is generated using the data of the measurement result and the estimated value of the position and orientation of the mobile robot obtained by the dead reckoning method. . Further, in the present invention, it is assumed that the walls existing in the indoor passage environment generally have a positional relationship orthogonal to each other. Therefore, the object of the present invention is to extract information of a wall that roughly represents the shape of the passage environment as line segments orthogonal to each other, and to generate a two-dimensional map using the extracted line segments. Hereinafter, a method of generating a map according to the present invention will be described step by step with reference to FIG.

(Step 101) 1 by laser sensor
In the shape measurement for the first time, it is necessary to perform a state in which the direction of the mobile robot is substantially equal to the wall existing in parallel with the traveling direction. Therefore, the initial position and orientation of the robot are appropriately set before starting the map generation. However, any posture can be taken for the second and subsequent shape measurements.

(Step 102) 1 by laser sensor
The first shape measurement is performed after the processing in step 101 is correctly completed. The second and subsequent shape measurements are performed after the mobile robot is guided to the next shape measurement position by a joystick or the like and is stopped. The data that can be obtained as a result of the shape measurement are a plurality of point coordinates on the surface of a wall or an obstacle existing in the passage environment, using the position and orientation of the laser sensor as the coordinate origin, and an appropriately provided buffer. Store in.

(Step 103) The data obtained as a result of the processing in step 102 is a set of point data having the position and orientation of the laser sensor as the coordinate origin, and is expressed in a so-called local coordinate system. Therefore, for the subsequent processing, using the displacement of the position and the direction from the state of the previous shape measurement of the mobile robot estimated by the dead reckoning method, the shape measurement data in the global coordinate system (point data Set) and stores the result in a suitably provided buffer.

(Step 104) Grit map data is created by dividing the shape measurement data in the global coordinate system obtained as a result of the processing of step 103 into a grid in the X-axis direction and the Y-axis direction. That is, a buffer represented by A [j] [i] (j is 1 to n and i is an integer of 1 to m) is secured for all the storage areas, and the area A [j]
In [i], the number of all point data satisfying the condition of the following expression and respective coordinate values are stored. ( _Xmax / m) * (i-1) <x <( _xmax / m) * i ( _ymax / n) * (j-1) <y <( _ymax / n) * j x _max and y _max are the maximum values of the defined grid map data in the X-axis direction and the Y-axis direction, respectively. Further, the number of stored all-point data is set to an appropriately set number K.
If the number is greater than or equal to the number, the state flag attached to each grid map data is turned on, and otherwise the state flag is turned off.

(Step 105) Grid map data A [j] [i] (j obtained as a result of the processing of step 104
Are 1 to n and i is an integer of 1 to m), and there are K or more stored point data, and the target grid map data A
[J] [i] If the surrounding grid map data satisfies at least one of the following four conditions, the state flag attached to the target grid map data is turned off. The purpose of this processing is to remove data of the grid map that includes points forming the intersection of the X-axis direction and the Y-axis direction in the grid map data. Number of point data of A [j + 1] [i] ≧ K∩A [j] [i−
1] Number of point data ≧ K A [j + 1] [i] Number of point data ≧ K∩A [j] [i +
1] Number of point data ≧ K A [j−1] [i] Number of point data ≧ K∩A [j] [i−
1] Number of point data ≧ K A [j−1] [i] Number of point data ≧ K∩A [j] [i +
1] Number of point data ≧ K

(Step 106) A leveling process is performed on grid map data for which the state flag obtained from the process of step 105 is on, and grid data storing point data constituting only the same line segment is stored. Each combination of map data (hereinafter referred to as a grid map data sequence) is generated.

(Step 107) Among the data strings of the grid map obtained as a result of the processing in step 106, the data strings whose number of grids constituting each is smaller than a suitably set threshold value are set in this grid map data. The column is excluded from the target of the subsequent line segment extraction processing. As a result, the grid map data sequence to be subjected to the subsequent line segment extraction processing is only the one that can form a relatively long line segment (hereinafter referred to as a main line segment).

(Step 108) The least squares approximation method is applied to all point data in each grid map data sequence which can form a main line segment obtained as a result of the processing of Step 107,
Approximate by one straight line. At this time, if the evaluation value of the poor approximation obtained by performing the approximation processing (hereinafter, referred to as a least square error) becomes smaller than a threshold value appropriately set in advance, a straight line obtained as a result is obtained. (Coordinates of the inclination and the contact piece) are stored in the appropriately profited buffers. In addition, in order to determine the coordinate values corresponding to the two end points of each main line, two points which are the farthest apart from each other are extracted from the data of the points used in the straight line approximation processing, and each point is approximated. Find the corresponding coordinates on the straight line. Then, the coordinate values of the two end points obtained as a result are stored in an appropriate area in the buffer in which the parameters representing the corresponding straight lines are stored.

(Step 109) Using the processing result of step 108 for the N-th shape measurement data, that is, the parameter representing the main line segment and the extraction result of the main line segment based on the (N-1) th shape measurement data And the error of the estimated value of the position and orientation of the mobile robot by the dead reckoning method. The processing of step 109 will be described in detail in steps 201 to 206 of the flowchart of FIG.

(Step 110) Using the error of the estimated value of the position and orientation of the mobile robot by the dead reckoning method obtained as a result of the processing of step 109, the global coordinates are appropriately corrected so that the error of the estimated value is appropriately corrected. The coordinate conversion process is performed on the N-th shape measurement data in the system, and the (N-1) -th and N-th shape measurement data are merged to perform a line segment extraction process again. The same coordinate transformation processing as described above is also performed on the extracted main line segment, and the information on the corrected main line segment obtained as a result is stored in an appropriately provided buffer, and the (N + 1) -th time Used for main line segment extraction processing.

(Step 111) Using a man-machine interface device mounted on the mobile robot used for shape measurement, the operator informs the mobile robot of the end of the map generation processing as necessary, and the transmitted contents Determines whether to end or continue the map generation processing in accordance with.

(Step 112) After completing the main line segment extraction processing for the N-th shape measurement data, the (N + 1) -th
In order to obtain the second shape measurement data, the position and orientation of the mobile robot equipped with the laser sensor in the passage environment are updated. This update is performed by an operator of the mobile robot guiding the mobile robot to an appropriate place using a joystick or the like. Next, the details of the processing in step 109 will be described for each of steps 201 to 206 in the flowchart of FIG.

(Step 201) Each main line segment obtained as a result of the processing of step 108 for the N-th shape measurement data, and the main line obtained as a result of the main line segment extraction processing based on the (N-1) th shape measurement data Determine the response in minutes. The condition for determining the correspondence is that the difference between the inclinations, which are the parameters representing the main line segments, and the difference between the coordinate values in the X-axis direction and the Y-axis direction of the center point of the two end points are set appropriately. It is assumed that it is smaller than the threshold value. Here, a distance difference in the Y-axis direction is used for a main line extending in the X-axis direction, and a distance difference in the X-axis direction is used for a main line extending in the Y-axis direction. Then, the parameters representing the main line segments determined to correspond to each other are stored in appropriately provided buffers.

(Step 202) The difference between the directions between the main lines determined in the course of the processing in step 201 is stored in an appropriately provided buffer.

(Step 203) An average value of the difference in direction between the main lines determined as a result of the processing in step 202 is obtained, and this value is defined as Δθ to be an error in the estimated value of the direction of the mobile robot. Store in an appropriately provided buffer.

(Step 204) Using the Δθ obtained as a result of the processing in step 203, the effect of the error in the estimated value of the orientation of the mobile robot is removed. In other words, the rotation transformation of −Δθ is performed on all the main line segments obtained by the N-th main line segment extraction process around the origin of the local coordinate system.

(Step 205) With respect to the main line segment extending in the X-axis direction that can be determined to correspond to the main line segment from which the error of the estimated value of the direction obtained as a result of the processing in step 204 has been removed, the distance in the Y-axis direction is obtained. A distance difference in the X-axis direction is calculated for the main line extending in the Y-axis direction, and stored in an appropriately provided buffer so as to recognize the distance difference in any direction using a flag or the like. Here, the distance difference is a difference between coordinate values in the X-axis direction or the Y-axis direction of the center point of the two end points of each main line.

(Step 206) The average value of all the distance differences in the X-axis direction and the average value of all the distance differences in the Y-axis direction obtained as a result of the processing of Step 205 are obtained, and the respective results are calculated by the dead reckoning method. Error of the estimated value of the mobile robot in the X-axis direction ΔX or error of the estimated value in the Y-axis direction ΔY
And store it in an appropriately provided buffer. By performing the map generation by the above-described method, the error of the estimated value of the position and the orientation of the mobile robot by the dead reckoning method can be gradually cleared, and the shape of the straight wall is expressed in the XY rectangular coordinate system. In the above, it can be easily extracted as a straight line component.

[0023]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, specific embodiments of the present invention will be described. FIG. 3 shows an example of the appearance of a mobile robot equipped with a rotary scanning laser sensor used in the present invention.
Reference numeral 1 denotes a rotary scanning laser sensor; 302, a central control unit of a map generation system; 303, a cable for transmitting a command signal from a joystick to a mobile robot;
Four joysticks, 305 is a mobile robot, 306 is a drive wheel with an encoder, and 307 is a caster. FIG. 4 is an example of a system configuration of hardware of a mobile robot equipped with a rotary scanning type laser sensor used in the present invention. Reference numeral 401 denotes the joystick 304;
402 is a cable 303 for transmitting a command signal from the joystick 304, 403 is an arithmetic unit to the central control unit 302, and 404 and 405 are arithmetic units 403
A bus connecting the sensor driver of the laser sensor 301 to the sensor driver 406; a sensor driver 406 of the laser sensor 301; a cable 407 for transmitting a sensor signal from the sensor head of the laser sensor 301 to the sensor driver 406; 408, a sensor head unit of the laser sensor; 409, 411, a bus connecting the arithmetic unit 403 and the storage unit 410 of the central control unit 302;
Are the storage units of the central control unit 302, 412 and 413 are buses connecting the arithmetic unit 403 and the motor driver of the driving wheel 306, 414 is the motor driver of the driving wheel 306, 415 and 416 are the driving wheel 306. A cable 417 for transmitting signals between the motor drivers 414;
Indicates the driving wheels.

FIG. 5 shows an example of an indoor passage environment to which the map generation method of the present invention is applied.
Reference numeral 07 denotes a single wall, and reference numerals 503, 504, and 505 denote positions where the mobile robot performs shape measurement using the laser sensor 301, respectively. However, shape measurement is performed in the order of 503, 504, and 505. FIG. 6 shows the data obtained by performing the first shape measurement at the position 503 in the local coordinate system of the horizontal axis X and the vertical axis Y, and 601 in the figure indicates measurement data of the wall 502. ,
602 is measurement data of the wall 501, 604 is the wall 5
Measurement data 07 and 603 indicate the origin of the local coordinate system. FIG. 7 shows the data obtained by performing the second shape measurement at the position 504 in the local coordinate system of the horizontal axis X and the vertical axis Y, and 701 in FIG.
02 measurement data, 702 indicates the measurement data of the wall 501, 704 indicates the measurement data of the wall 506, 705 indicates the measurement data of the wall 507, and 703 indicates the origin of the local coordinate system. FIG. 8 shows the data obtained by performing the third shape measurement at the position 505 in the local coordinate system of the horizontal axis X and the vertical axis Y, and 801 in the figure indicates measurement data of the wall 501. , 803 indicate measurement data of the wall 506, and 802 indicates the origin of the local coordinate system.
In order to make the following description general and easy, the first
Assuming that the process of extracting the main line of the data obtained by performing the second shape measurement has been correctly completed, the process of extracting the main line of the data obtained by performing the second shape measurement will be described below as an example. . The data obtained by performing the second shape measurement by the process of step 103 is re-expressed in the global coordinate system, and the number of grids in the X-axis direction used in the process of step 104 is set to 12
(M = 12), the number of grids in the Y-axis direction is 12 (y = 1
FIG. 9 shows the result of projection onto the image generated as 2), in which 901 is a grid represented by A [1] [1], 902 is a grid represented by A [12] [1],
Reference numeral 903 denotes a grid represented by A [1] [12];
4 is a grid represented by A [12] [12], 90
5, 906, 907, and 908 indicate data obtained as a result of coordinate conversion of the measurement data 701, 702, 704, and 705, respectively. In the process of step 104, the number of point data stored in each grid is one or more (K =
FIG. 10 shows the result of generating grid map data of the shape measurement data shown in FIG. 9 by turning on the state flag attached to each grid map data in the case of 1).
1001, 1002, 1003, and 1004 in the figure are measurement data 905, 906, and 90 after the coordinate transformation, respectively.
7 and 908 show grids whose status flags have been turned on. In the processing of step 105,
FIG. 11 shows the result of removing the grid map data including the point data forming the intersection between the X-axis direction and the Y-axis direction and generating a grid map data string for each line of the labeling processing in step 106. Yes, 1102 in the figure,
1104 is grid map data 1101, 1103, 1105, which stores the point data to be removed.
1106 is the wall 502, 501, 506, 5
7 shows a grid map data sequence constituting only the point data corresponding to 07. In the processing of the step 107, the grid map data strings 1101, 1103, 110
5 and 1106 having three or less grids are excluded from the target of the subsequent line segment extraction processing, and the results obtained are shown in FIG.
Are grid map data strings remaining as targets of the subsequent main line extraction processing. In the processing of the step 108, a straight line approximation by the least squares approximation method is performed on all the point data in the grid map data examples 1201 and 1202, and the least square error obtained at that time is smaller than the appropriately set threshold value. The main line segments that have become smaller and have been further subjected to the extraction processing of the two end points are denoted by 1301 and 1304 in FIG.
This is the line segment indicated by. Further, 1302, 1 in FIG.
Reference numerals 303 and 1305 denote results obtained by performing the main line segment extraction processing on the first shape measurement data, and are line segments corresponding to the walls 502, 501, and 507, respectively.
In the processing of step 201, the main line segments 1301 and 1
302 can be determined, and the main line segments 1304 and 1303 can be determined.
Are determined to correspond, the steps 202, 20
In the processes of Steps 3 and 204, the main line segments 1301 and 1304 are subjected to coordinate transformation for rotation about the origin of the local coordinate system of the second shape measurement position 504, and the estimated value of the orientation of the mobile robot by the dead reckoning method is calculated. FIG. 14 shows the result of removing the influence of the error. In the figure, reference numerals 1401 and 1402 denote main line segments 1301 and 1304 after coordinate conversion before the coordinate conversion, respectively. Step 20
In steps 5 and 206, the error (ΔX) of the estimated value of the mobile robot in the X-axis direction by the dead reckoning method is obtained from the difference between the coordinate values of the main line segments 1401 and 1302 in the X-axis direction.
The error (ΔY) of the estimated value of the mobile robot in the Y-axis direction by the dead reckoning method is obtained from the difference between the coordinate values of the main line segments 1402 and 1303 in the Y-axis direction. The first shape measurement data shown in FIG. 6 by the coordinate conversion process using
In the line segment extraction process of the data obtained by fusing the second shape measurement data shown in FIG. 7, the result of performing the new line segment extraction process is shown by the main line segments 1501, 1502, 1 in FIG.
503. In the process of step 111, when the operator notifies the mobile robot of the end of the map generation process using the joystick, the mobile robot ends the map generation process by determining this, and In step 112, the mobile robot is guided to the third shape measurement position 505, and then the above-described line segment extraction processing is performed again.

[0025]

As described above, according to the present invention, even if an error occurs in the estimated value of the position and orientation of the robot by the dead reckoning method, the error is cleared if the error is relatively small. This makes it possible to easily perform line segment extraction processing on the result of shape measurement.

[Brief description of the drawings]

FIG. 1 is a main flowchart of the present invention.

FIG. 2 is a sub flowchart of the present invention.

FIG. 3 is a diagram illustrating an example of the appearance of a mobile robot that implements the present invention.

FIG. 4 is a diagram illustrating an example of a hardware configuration of a mobile robot that implements the present invention.

FIG. 5 is a diagram showing an example of an indoor passage environment to which the present invention is applied;

FIG. 6 is a view for explaining a line segment extraction algorithm according to the present invention;

FIG. 7 is a view for explaining a line segment extraction algorithm according to the present invention;

FIG. 8 is a view for explaining a line segment extraction algorithm according to the present invention;

FIG. 9 is a view for explaining a line segment extraction algorithm according to the present invention;

FIG. 10 is a diagram for explaining an algorithm of line segment extraction according to the present invention.

FIG. 11 is a diagram for explaining an algorithm of line segment extraction according to the present invention.

FIG. 12 is a view for explaining a line segment extraction algorithm according to the present invention;

FIG. 13 is a diagram illustrating an algorithm for extracting a line segment according to the present invention.

FIG. 14 is a view for explaining a line segment extraction algorithm according to the present invention;

FIG. 15 is a view for explaining a line segment extraction algorithm according to the present invention;

[Explanation of symbols]

301 laser sensor 302 central control unit 304, 401 joystick 305 mobile robot 306, 417 drive wheel with encoder 307 caster 403 arithmetic unit 404, 405, 409, 411, 412, 413 bus 406 laser sensor driver 303, 402, 407, 415, 416 Cable 408 Laser sensor head unit 410 Storage unit 414 Motor driver

Claims (2)

[Claims] 1. A measurement result of a shape of a passage environment in which a wall having an orthogonal positional relationship exists by a scanning laser sensor mounted on a mobile robot, and an estimation of a position and an orientation of the mobile robot obtained by a dead reckoning method. Generating a map of the passage environment required by the mobile robot by extracting, from the values, line segments corresponding to mutually orthogonal walls that schematically represent the shape of the passage environment; Map generation method. 2. Estimating the position and orientation of a mobile robot by a dead reckoning method using shape measurement data in a local coordinate system with the position and orientation of the laser sensor obtained using the scanning laser sensor as a coordinate origin. Using the values, the grid map data is re-expressed in a global coordinate system by a coordinate transformation process to generate grid map data, and grid map data including point data forming an intersection in the X-axis direction and the Y-axis direction in the grid map data. After the removal, a grid map data sequence to be described later is generated by a labeling process.If the number of grids in each grid map data sequence is smaller than a preset threshold, the grid map data sequence is excluded from the target of the subsequent line segment extraction process. Exclude the grid map data sequence that can compose only relatively long line segments and extract all points data included in each remaining grid map data sequence. For the data, the least squares error approximation by a straight line is performed, and when the least squares error that is the evaluation value is larger than a preset threshold, it is excluded from the target of the subsequent line segment extraction processing, and the reliability is as high as possible. Only the straight line is treated in the subsequent line segment extraction processing, and the corresponding point on the approximate straight line of the two end points where the mutual distance between all the point data used as the input of the straight line approximation processing is the longest is calculated. From the result of the matching process between the line segment obtained at this time and the position and direction of the line segment obtained by processing the previous shape measurement data, the error of the estimated value of the position and direction of the mobile robot by the dead reckoning method 2. The path environment map generating method according to claim 1, wherein a line segment is extracted by removing the influence of the error by a coordinate transformation process.