JP5130419B2 - Self-position recognition method and self-position recognition device - Google Patents

Description

  The present invention relates to a method and apparatus suitable for a moving body such as a robot or a vehicle that runs unattended to recognize its own position.

  In recent years, a wide variety of robots are active in various situations. Above all, robots that can work while recognizing their own position while moving are expected to help humans in factories, offices, hospitals, etc., and their research is actively conducted . Currently, various self-position recognition methods have been proposed for robots. A method generally used in a building is a method of recognizing a self-position by accumulating the amount of movement of wheels. As is well known, this method, which is a type of dead reconning method, has a problem that the accumulated error becomes large.

Therefore, a method has been proposed in which a landmark is artificially installed in the environment and the self-position is recognized based on the landmark (for example, Non-Patent Document 1).
In addition, a method is known in which a magnetic marker made of, for example, a magnetic tape is installed on the movement path, a sensor for detecting the magnetic marker is provided in the robot, and the self-position is recognized by detecting the magnetic marker by the sensor. And the traveling control method which combined the dead reconning method and the method of detecting a magnetic marker is proposed (patent document 1).

Kazumasa Takeuchi, Jun Ohta, Kazuo Ikeda, Yasumichi Aiyama, Tamio Arai “Mobile Robot Navigation Using Artificial Landmarks”, Japan Society of Mechanical Engineers [No. 99-9] Robotics and Mechatronics Lecture ′99 Proceedings 1P1-40-052 JP 2002-73171 A

However, there are cases where landmarks have problems in their installation. In addition, the burden of work and cost is great for installing landmarks.
On the other hand, in the method of installing a magnetic marker, the magnetic field distribution of the environment must be obtained in advance and the magnetic marker must be installed in a place where there is little magnetic noise. Therefore, also in this method, the burden concerning the operation | work which installs a magnetic marker is large. This burden also arises when the travel route is changed. Moreover, although a magnetic marker is generally installed on the floor, it may be aesthetically undesirable to install a magnetic marker on the floor surface.
SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to provide a method and apparatus capable of recognizing a self-position without installing a landmark, a magnetic marker or the like.

  Ferromagnetic materials such as steel members and steel nodes that make up buildings, iron members that are components of furniture, electrical appliances, etc. installed in buildings are inevitably magnetized in the process of manufacturing and construction, causing residual magnetism. I am tinged. This residual magnetism is direct current magnetism whose strength hardly varies with time (Non-Patent Document 2). The present invention is based on the premise that magnetism is used as a marker for self-position recognition, but it does not use magnetism generated by newly installing a magnetic marker in the environment, but depends on the environment as described above. Use the generated magnetism.

Toshinfumi Shinno “Case Study of Ambient Magnetic Field Measurement Method (Part 3)”, Summary of Academic Lectures of Architectural Institute of Japan, September 1996

In the present invention, magnetism depending on an environment in which a moving body such as a robot moves and a position where the magnetism is measured are stored in association with each other. And this invention recognizes a self-position by comparing the magnetism-position information which is matched information, and the magnetism measured when recognizing a self-position.
That is, the method of the present invention for the moving body to recognize its own position measures the reference magnetism in the environment in which the moving body moves, and the magnetism-position in which the reference magnetism is associated with the position where the reference magnetism is measured. A step (a) for storing information in advance, a step (b) for measuring the measured magnetism by a magnetic sensor provided in the moving body in the environment, a reference magnetism of the magnetism-position information stored in the step (a), (C) specifying the position where the measured magnetism is measured from the magnetism-position information by comparing the measured magnetism measured in step (b) with the magnetism-position information, and axis, a reference magnetic strength along the three axes of the Y-axis and Z-axis and the position is characterized der Rukoto those associated.
Since the self-position recognition method of the present invention can recognize the self-position by using magnetism that depends on the environment, there is no need to newly install a magnetic marker.
Here, when comparing the magnetic-position information in which the intensity and position of the reference magnetism along one axis are associated with the measured magnetism, if the change in the reference magnetism is small, the measured magnetism is calculated from the magnet-position information. It may be difficult to identify the position where the is measured. However, according to the self-position recognition method of the present invention, the magnetism-position information is obtained by associating the strength and position of the reference magnetism along three orthogonal axes, the X axis, the Y axis, and the Z axis. Therefore, if the change in the intensity of the reference magnetism along any one of the axes is large, there is an effect that it is easy to specify the position where the measured magnetism is measured from the magnetism-position information.

  In the self-position recognition method of the present invention, when the moving body moves in the environment of the building, the reference magnetism and the actually measured magnetism have the ferromagnetic body constituting the building and the ferromagnetic body installed in the building. Measured based on magnetism. As mentioned above, ferromagnetic materials such as steel members and steel nodes that make up buildings, iron members that are components of furniture, appliances, and electrical equipment are inevitably worn during the manufacturing and construction process. It is magnetized and has residual magnetism. Based on this residual magnetism, the magnetism generated in the environment is measured and used as the reference magnetism and the actually measured magnetism.

In step (c) of the self-position recognition method of the present invention, the magnetic-position information and the measured magnetism are compared by linearly matching the reference magnetism data string of the magnet-position information and the measured magnetism data string. It is preferable. This is because the position where the measured magnetism is measured can be accurately identified from the magnetism-position information.

The linear matching is preferably performed according to the following procedure.
Each of the magnetic-position information data strings regarding the X axis, the Y axis, and the Z axis is linearly matched with each of the measured magnetic data strings regarding the X axis, the Y axis, and the Z axis. However, in any of the X-axis to Z-axis, there may be a case where a matching portion cannot be specified between the magnetic-position information data string and the measured magnetism data string. In that case, the measurement of the reference magnetism is continued while moving the moving body, and the linear matching process is performed using the data example including the newly measured reference magnetism.

  In the self-position recognition method of the present invention, it is preferable that the magnetism-position information is based on reference magnetism measured along a predetermined path set in the environment. This is because the time required for comparing the reference magnetic data string and the measured magnetic data string is shortened by reducing the number of stored magnetic-position information data. On the other hand, when moving the robot, if the robot can be moved along a predetermined route, the purpose of moving the robot is usually achieved.

  As an apparatus for implementing the self-position recognition method according to the present invention, storage means for storing magnetism-position information in which a reference magnetism measured in an environment in which a moving body moves and a position where the reference magnetism is measured are associated with each other; The position where the measured magnetism is measured is identified by comparing the magnetic sensor that measures the measured magnetism in the environment, the reference magnetism of the magnetism-position information stored in the storage means, and the measured magnetism measured by the magnetic sensor. The present invention proposes a self-position recognizing device comprising position specifying means.

  According to the present invention, since the self-position can be recognized using the magnetism originally generated depending on the environment, there is no need to newly install a magnetic marker, and the magnetic marker is re-applied even when the movement path is changed. There is no need to perform installation work. Moreover, since it is not necessary to install a magnetic marker on the floor, it does not harm the aesthetics of the environment.

Hereinafter, the present invention will be described in detail based on embodiments shown in the accompanying drawings.
FIG. 1 is a side view showing a schematic configuration of a wheel-type mobile robot (mobile body, hereinafter simply referred to as a robot) 1 according to the present embodiment.
The robot 1 includes a pair of left and right drive wheels 3 and 4 at the center lower portion in the front-rear direction of the box-shaped main body 2, and auxiliary wheels 5 at substantially lower corners of the main body 2. Further, a magnetic / gyro sensor 20 and a CCD camera 30 are installed on the front side (right side in the figure) of the main body 2. The magnetic / gyro sensor 20 and the CCD camera 30 are electrically connected to the control unit 10 installed inside the main body 2.
The magnetic / gyro sensor 20 can measure DC magnetism (hereinafter simply referred to as magnetism) along three axes orthogonal to each other. Here, the three axes include the X axis along the traveling direction of the robot 1, the Y axis parallel to the floor surface and perpendicular to the X axis, and the direction parallel to the vertical direction and perpendicular to the X axis as the Z axis. The magnetism to be measured is used for creation of magnetic-position information and self-position recognition described later. In addition to magnetism, the magnetic / gyro sensor 20 can measure angular velocity. The angular velocity can be used for straight running control of the robot. In the present embodiment, an example using the magnetic / gyro sensor 20 in which the magnetic sensor and the gyro sensor are integrated is shown, but each may be provided separately.

FIG. 2 is a block diagram showing the configuration of the control system of the robot 1.
In FIG. 2, the drive wheels 3 and 4 are rotationally driven by drive motors 6 and 7 via reduction gears (not shown). The drive motors 6 and 7 are respectively provided with encoders 8 and 9 for detecting the rotation speed (rotation speed) of the drive wheels 3 and 4. Further, the drive motors 6 and 7 are independently driven and controlled by the travel controller 11 via the motor drive circuit 12. The travel controller 11 can change the orientation of the robot 1 by rotating the drive motors 6 and 7 at different rotational speeds.

The travel controller 11 is composed of, for example, a computer including a CPU, ROM, RAM, and an input / output circuit, and has a position specifying unit 11a and a storage unit 11b.
The travel controller 11 stores the magnetism (reference magnetism) measured by the magnetism / gyro sensor 20 and the position where the magnetism is measured in the storage unit 11b in association with each other. The associated data (magnetism-position information) is stored in advance in the storage means 11b before the robot 1 recognizes its own position. The storage unit 11b stores programs related to various controls performed by the travel controller 11.
The position specifying means 11a of the travel controller 11 acquires the magnetism (measured magnetism) measured by the magnetism / gyro sensor 20 when the robot 1 recognizes its own position, while the magnetism stored in the storage means 11b. Read location information. The position specifying unit 11b specifies the position where the actually measured magnetism is measured by comparing the read reference magnetism with the acquired actually measured magnetism.
The travel controller 11 recognizes its own position by a known dead reconning method based on the rotational speed of the drive wheels 3 and 4 detected by the encoders 8 and 9, and the robot 1 moves along a predetermined path. The drive motors 6 and 7 are controlled as described above, but since this is a known technique, the details are omitted.

FIG. 3 shows an example of the data structure of magnetic-position information.
As shown in FIG. 3, in the magnetism-position information, an arbitrary position on the two-dimensional coordinates (x, y) and the magnetism (reference magnetism) at the position are associated with each other. That is, the reference magnetism measured at the position (x ₁ , y ₁ ) is Magx ₁ , Magy ₁ , Magz ₁ , and the reference magnetism measured at the position (x ₂ , y ₂ ) is Magx ₂ , FIG. 3 shows that it is Magy ₂ and Magz ₂ . Incidentally, MAGX ₁ magnetic strength along the X-axis magnetic gyro sensor 20, Magy ₁ is the intensity of the magnetic along the Y-axis magnetic gyro sensor 20, Magz ₁ in the Z-axis magnetic gyro sensor 20 It shows the magnetic strength along.

  In the self-position recognition method based on magnetism-position information, it is desirable that magnetic strength and position coordinates are stored in association with each other at any position in the movement environment of the robot 1, and the position coordinates can be recognized in real time based on this. However, this method is not practical at the current technical level. This is because the larger the movement environment of the robot, the larger the number of data to be stored, and the processing for recognizing the position coordinates (for example, linear matching processing) takes a considerable time. Therefore, in order to be able to recognize the self-position more easily and earlier, in the present embodiment, the magnetic-position information is stored according to the procedure shown in FIG. That is, the movement path of the robot 1 is determined in the movement space of the robot 1 (S101 in FIG. 4). While the robot 1 travels straight ahead at a constant speed (S103 in FIG. 4) from the origin to the end point of the determined moving path, magnetism (DC magnetism) is measured by the magnetic / gyro sensor 20 (S105 in FIG. 4). At the same time, the moving distance from the origin is measured by the encoders 8 and 9 (S107 in FIG. 4). The magnetic intensity as the output from the magnetic / gyro sensor 20 and the position coordinates specified by the movement distance as the outputs from the encoders 8 and 9 are associated with each other and stored in the memory 11b (S109 in FIG. 4).

An example of magnetic-position information storage processing actually performed according to the above procedure will be described below.
As shown in FIG. 5, there are two routes (y1, y2) in the y direction and seven routes (x1, x2, x3, x4, x5, 1 m from the start point (S) in the y direction in the x direction, as shown in FIG. x6, x7). The robot 1 was moved on all moving paths, and the magnetic-position information was stored. 6 (detected by the X axis of the magnetic / gyro sensor 20), FIG. 7 (detected by the Y axis of the magnetic / gyro sensor 20), and FIG. 8 (magnetic / gyro). (Detected by the Z axis of the sensor 20). 6, 7, and 8, the vertical axis indicates the DC magnetic strength, and the xy plane indicates the position coordinates.

As will be described in detail later, the present embodiment linearly matches the reference magnetism data string of magnetism-position information and the measured magnetism data string detected by the magnetism / gyro sensor 20, and the position associated with the reference magnetism. By specifying the coordinates, the self position is recognized. Therefore, it can be said that the position recognition is easier as the change in the intensity of the actually measured magnetism is larger than the movement distance of the robot.
With this in mind, referring to FIGS. 6 to 8, all of FIGS. 6 to 8 have large changes in the reference magnetism along the path (y1, y2) in the y direction. In the case of a path in the y direction, any of the magnetic-position information obtained via the X axis (FIG. 6A), the Y axis (FIG. 7A), and the Z axis (FIG. 8A) If one of them is used and compared with the measured magnetism, self-position recognition is possible. Of course, all three magnetism-position information can be used, but it takes time to compare (linear matching) with the measured magnetism.
On the other hand, referring to FIGS. 6 to 8, the change in the reference magnetism in the path in the x direction is particularly small in FIGS. In such a case, it is necessary to compare the magnetism-position information corresponding to the three axes (X axis, Y axis, Z axis) of the magnetic / gyro sensor 20 with the measured magnetism.

Next, the procedure in which the robot 1 recognizes its own position using the magnetic-position information will be described with reference to FIG. This procedure is based on the premise that the magnetic-position information is stored in advance in the storage unit 11b (S401 in FIG. 9). Further, the procedure shown in FIG. 9 assumes that the robot 1 moving on a path surrounded by walls on both sides recognizes its own position.
When performing self-position recognition, an image around the passage is taken by the CCD camera 30. The position specifying unit 11a captures the captured image (S201 in FIG. 9). The position specifying unit 11a processes the captured image and extracts an edge straight line corresponding to the boundary between the wall and the floor of the passage using Hough transform (S203 in FIG. 9). The position specifying unit 11a specifies the distance (ρ) from the edge straight line to the robot 1 (CCD camera 30). This distance is obtained in the process of extracting the edge straight line by the Hough transform.

The Hough transform is a technique for extracting a parametric model from an image or the like. In the Hough transform, a parameter is estimated by selecting a parameter that is significantly higher from each point in all parameter voting spaces in which the point may exist.
The principle of straight line (edge straight line) detection by Hough transform is as follows.
In the case of straight line detection, as shown in FIG. 10A, all straight lines passing through the point (x, y) on the rectangular coordinates are the angle θ and length (distance) between the straight line and the x axis of the straight line. ) Ρ. If (rho) is calculated | required from FIG.10 (b), it will become following Formula (1).
ρ = x cos θ + ysin θ (1)
Expression (1) is obtained by converting a point (x, y) on a rectangular coordinate into a polar coordinate two-dimensional space having an angle θ and a distance ρ. When the number is added for each angle θ and distance ρ, the combination of the largest number (angle θ and distance ρ) is returned to the original rectangular coordinates to form a set of points most likely to be straight. This collection of points is an edge straight line.

Next, using the measured distance (ρ) and the azimuth of the robot 1 detected from the magnetic / gyro sensor 20, the movement path closest to the robot 1 is selected (S207 in FIG. 9). This movement path is selected from a plurality determined when the magnetic-position information is stored.
Next, linear matching is performed (S209 in FIG. 9). Linear matching is performed by comparing the measured magnetism (FIG. 9 S301) measured on the selected path with the magnetic-position information (FIG. 9 S401) stored in advance for the path.

A linear matching method will be described with reference to FIG.
FIG. 11 shows an example of a data string of magnetic-position information corresponding to the selected path, where the vertical axis indicates the magnetic intensity and the horizontal axis indicates the position on the path. A curve indicated by a solid line in FIG. 11A corresponds to the X-axis magnetism of the magnetic / gyro sensor 20. Similarly, the curve (solid line) in FIG. 11B corresponds to the Y-axis magnetism, and the curve (solid line) in FIG. 11C corresponds to the Z-axis magnetism.
In FIGS. 11A to 11C, it is assumed that a curve indicated by a dotted line indicates a data string of actually measured magnetism measured by the magnetic / gyro sensor 20 at the time of position recognition.

The linear matching is a method for performing data matching by shifting the data sequence in a certain step from the beginning to the end of the data sequence. FIG. 11 shows one step, and specifies a portion where the form of the curve (data string) indicated by the dotted line matches (or approximates) the form of the curve (data string) indicated by the straight line (S211 in FIG. 9). ). In the case of FIG. 11, the matching part can be specified for the X axis and the Y axis, but the matching part may not be specified for the Z axis. This is because in the case of the Z-axis (FIG. 11C), since the change in magnetism along the path is small, there are a plurality of portions that are determined to coincide with the measured magnetic curve.
In the case of the example shown in FIG. 11, since the matching part can be specified for the X-axis and the Y-axis, the position coordinate associated with the last data in the matching data string is read (S213 in FIG. 9). Is identified as the current position of the robot 1 (S215 in FIG. 9).
When a matching part cannot be specified for any of the X-axis to Z-axis, the magnetism / gyro sensor 20 continues to measure magnetism while moving the robot 1 (S301 in FIG. 9). Then, the matching process is performed in the same manner as described above using the data example including the newly measured magnetism and shifting the steps.

According to the present embodiment, it is possible to recognize the self position regardless of the position without performing the position initialization with respect to the robot 1.
Further, according to the present embodiment, since linear matching is performed for each of the actually measured magnetism detected by the three axes of the magnetic / gyro sensor 20, there is a high possibility that mismatching, that is, a matching portion cannot be specified. small. However, the present invention is not limited to a form in which linear matching is performed using all the reference magnetism measured in the three axes of the X axis, the Y axis, and the Z axis.

Next, an example of a method for moving the robot 1 based on the position recognition method described above will be described.
The movement method according to the present embodiment is basically performed by a known dead reconning method, but has a feature in that the movement is performed while correcting using the position recognized by the position recognition method described above. .
The paths indicated by the broken lines in FIG. 12 (six paths in the x direction and three paths in the y direction) are the paths on which the magnetic-position information is created, and the robot 1 moves from the point O to the target location G.
First, the robot 1 selects the middle path y2 extending in the y direction closest to the point O (self) that is the current position of the robot 1. As described above, the route y2 can be selected based on the detection of a straight line (edge straight line) by the Hough transform, the distance ρ, and the orientation of the robot 1.
After selecting the route y2, the robot 1 moves to the route y2. This movement may be performed based on a conventional dead reconning method based on the number of wheel rotations obtained by the encoders 8 and 9 without using the self-position recognition method according to the present embodiment.
The robot 1 that has moved to the route y2 moves to the point A on the route y2. The movement to the point A may be basically performed based on the conventional dead reconning method. However, since the conventional dead reconning method causes an accumulation error, if it moves a long distance, it will deviate from a predetermined path. Therefore, in the present embodiment, the robot moves to the point A while correcting the position of the robot 1 by performing self-position recognition according to the present embodiment for each distance where the accumulated error is allowable. When the robot 1 moves to the point A that is the intersection of the route y2 and the route x4, the robot 1 changes its direction and moves on the route x4 toward the point B. Similarly to the movement from the point A to the point A, the robot 1 may be moved while appropriately correcting the position of the robot 1.

If the robot 1 moves to the point B, the robot 1 moves to the target location G after changing its direction. This movement can be performed based on the conventional dead reconning method.
In the movement using only the conventional dead reconning method, there is a problem that the accumulated error becomes large. However, according to the moving method according to the present embodiment, the self-position recognized based on the magnetic-position information can be used as the initial position of the dead reconning method. This means that the accumulated error is reset. Therefore, it can be said that the accumulated error does not occur according to the moving method according to the present embodiment.

Next, a result of an experiment in which the robot 1 is actually moved on the route will be described.
This experiment was obtained by actually measuring the position (recognition position) recognized by the robot 1 by moving the robot 1 straight on the path provided in the path of 1.75 m wide provided with walls on both sides. An error with respect to the position (actual measurement position) of the obtained robot 1 is obtained.
The experiment used two paths along the longitudinal direction (y direction) of the passage and two paths along the width direction (x direction) of the passage. The results are shown in FIGS. FIGS. 13 and 14 show the results of experiments in which the robot 1 is moved along a path 0.5 m and 1.25 m away from x = 0 (wall). FIGS. 15 and 16 show the robots. The result of the experiment which moved 1 on the path | route 1m and 2m away from the point of y = 0 is shown.

In addition, the creation and matching processing of magnetic-position information in this experiment was performed as follows.
In the y direction, the moving speed of the robot 1 was 0.06 m / s, and the reference magnetism was measured at 5400 points on the 6.0 m path (number of data: 5400). In the x direction, the moving speed of the robot 1 was set to 0.03 m / s, and the reference magnetism was measured at 1500 points on the 0.75 m route (1500 data pieces). In this experiment, the number of data to be matched in one step was set to 100 in consideration of the processing speed of linear matching. Therefore, the magnetic-position information used for the one-step matching process corresponds to a distance of 6.0 m / 54 ≈ 0.11 m in the y direction and 0.75 / 15 ≈ 0.05 m in the x direction.

In FIG. 13A, FIG. 14A, FIG. 15A, and FIG. 16A, ◯ indicates the recognition position, and X indicates the actual measurement position.
FIG. 13B, FIG. 14B, FIG. 15B, and FIG. 16B show errors between the recognized position and the actually measured position. The error between the recognition position and the actual measurement position is 6.2 cm at the maximum in the result of FIG. 13B, 8.2 cm at the maximum in the result of FIG. 14B, and 6.5 cm at the maximum in the result of FIG. In the result of FIG. 16B, it was confirmed that the maximum was 7.8 cm. This error (maximum value) is smaller than the cumulative error caused by the dead reconning method.

  Next, as shown in FIG. 17, the robot 1 is started from the start point S and moved only one round along a rectangular path of 6 m in the y direction and 0.75 m in the x direction. An experiment was conducted to determine whether or not The experiment was performed by correcting the dead reconning method by position recognition according to the present embodiment (Example) and using only the conventional dead reconning method using the rotation speed of the wheels obtained by the encoders 8 and 9. The case of moving (conventional example) was performed. The results are shown in FIG.

  In FIG. 17, the solid black line indicates the movement path, the alternate long and short dash line indicates the result of the example, and the dotted line indicates the result of the conventional example. As shown in FIG. 17, in the conventional example, the robot 1 cannot return to the start point S. This is because the floor slips and the rotational deviation of the drive wheels 3 and 4 of the robot 1 occur, and the error increases cumulatively. On the other hand, according to the present embodiment, the robot can return almost to the start point S.

In the conventional example, the difference between the return point of the robot 1 and the start point S is 3 cm in the y direction and 17 cm in the x direction. If the robot 1 moves repeatedly on the same route, this deviation gradually increases. Therefore, according to the conventional example, it is difficult to make the robot 1 work autonomously while moving a long distance.
On the other hand, according to the present embodiment, an error occurs between the return point and the departure point, but since the self-position is sequentially recognized based on the magnetic-position information, this error is not accumulated. Based on the present embodiment, an experiment was performed in which the route shown in FIG. 17 was moved eight laps. It was 2 cm in the x direction. Therefore, by using the self-position recognition method according to the present embodiment, it is possible to make the robot 1 work while moving a long distance.

Although this Embodiment demonstrated above assumes the robot 1 which moves in a building, this invention is not limited to this. If the magnetism is generated and there is no or small fluctuation of the magnetism, the self-position recognition method of the present invention can be implemented even outdoors.
In the present embodiment described above, the example in which the robot 1 is moved in combination with the conventional dead reconning method using the rotation speed of the drive wheels 3 and 4 obtained by the encoders 8 and 9 has been described. The usage form of the position recognition method is not limited to this. It can be combined with a conventional dead reconning method using information obtained at other inner world centers and information obtained at the outer world center, and the movement of the robot 1 can be controlled using only the self-position recognition method of the present invention.

  In the present invention, linear matching can be performed by combining magnetic-position information corresponding to the three axes of the X-axis, Y-axis, and Z-axis of the magnetic / gyro sensor 20. For example, when the robot 1 moves from the point A to the point D, consider a case where the robot 1 passes through the point B and the point C. In this case, from point A to point B, linear matching processing is performed on the magnetic-position information corresponding to the X axis. From point B to point C, linear matching processing is performed on the magnetic-position information corresponding to the Y axis. Further, from point C to point D, linear matching processing is performed on the magnetic-position information corresponding to the Z axis.

Furthermore, in the present invention, not only linear matching but also a known method such as a matching method by dynamic programming (Dynamic Programming) (DP matching) or a more general pattern matching (pattern matching). It is also possible to identify the position where the measured magnetism is measured by comparing the reference magnetism of the information with the measured magnetism.
Furthermore, in the present invention, since the reference magnetism of the magnetism-position information and the actually measured magnetism are compared, in addition to specifying the position coordinates, the azimuth can also be specified. Therefore, the self-position in the present invention has a concept including an orientation.
In addition, the configuration can be added to or changed from the above-described embodiment without departing from the spirit of the present invention.

It is a side view which shows the structure of the mobile robot in this Embodiment. It is a block diagram which shows the electric structure of the mobile robot in this Embodiment. It is a figure which shows an example of the data structure of magnetic-position information. It is a flowchart which shows the procedure which memorize | stores magnetic-position information. It is a figure which shows an example of a movement path | route. It is a figure which shows an example of the magnetic map produced based on the direct current magnetism measured by the X-axis of the magnetic and gyro sensor. It is a figure which shows an example of the magnetic map produced based on the direct current magnetism measured by the Y-axis of the magnetic and gyro sensor. It is a figure which shows an example of the magnetic map produced based on the direct current magnetism measured by the Z-axis of the magnetism / gyro sensor. It is a flowchart which shows the procedure in which a mobile robot recognizes a self position using magnetism-position information. It is a figure for demonstrating the principle of the straight line detection by Hough conversion. It is a figure for demonstrating the method of linear matching. It is a figure which shows a mode that a mobile robot moves to a target based on magnetism-position information. Indicates the error between the position recognized by the robot (recognized position) and the position of the robot actually measured (measured position) when the robot travels straight on a route 0.5 m away from the point x = 0. FIG. An error between the position recognized by the robot (recognized position) and the position of the robot obtained by actual measurement (measured position) when the robot travels straight on a path 1.25 m away from the point x = 0 is shown. FIG. FIG. 6 is a diagram showing an error between a position recognized by the robot (recognized position) and a position of the robot obtained by actual measurement (actually measured position) when the robot travels straight on a route 1 m away from a point where y = 0. is there. FIG. 5 is a diagram showing an error between a position recognized by the robot (recognized position) and a position of the robot obtained by actual measurement (measured position) when the robot travels straight on a path 2 m away from a point where y = 0. is there. It is a figure which shows the result when a robot is started from the start point S and moved only one round along a rectangular route.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Mobile robot, 2 ... Main body, 3, 4 ... Drive wheel, 5 ... Auxiliary wheel, 6, 7 ... Drive motor, 8, 9 ... Encoder, 10 ... Control part, 11 ... Travel controller, 12 ... Motor drive circuit, 20 ... Magnetic / gyro sensor, 30 ... CCD camera

Claims (6)

A method for a mobile object to recognize its own position,
(A) preliminarily storing magnetism-position information in which reference magnetism is measured in an environment in which the moving body moves and the reference magnetism and the position where the reference magnetism is measured are associated with each other;
(B) measuring actual magnetism by a magnetic sensor provided in the moving body in the environment;
A step of identifying the position where the measured magnetism is measured by comparing the reference magnetism of the magnetism-position information stored in the step (a) and the measured magnetism measured in the step (b). (C),
Equipped with a,
The magnetic - location information, the self-position to X-axis, Y-axis and wherein der Rukoto what the strength and the location of the reference magnetic is associated along three axes of Z-axis respectively orthogonal Recognition method. The moving body moves in the environment in a building,
2. The self-position according to claim 1, wherein the reference magnetism and the actually measured magnetism are based on magnetism included in a ferromagnetic body constituting the building and a ferromagnetic body installed in the building. Recognition method. The step (c) performs a comparison between the magnetic-position information and the measured magnetism by linearly matching the magnetic-position information data string and the measured magnetism data string. Item 3. The self-position recognition method according to item 1 or 2 . Linearly matching each of the data sequences of the magnetic-position information with respect to the X-axis, Y-axis, and Z-axis and each of the measured magnetic data sequences with respect to the X-axis, Y-axis, and Z-axis,
For any of the X-axis to Z-axis, when it is not possible to identify a matching part between the data string of the magnetic-position information and the data string of the measured magnetism,
The measurement of the reference magnetism is continued while moving the moving body, and the linear matching process is performed using the data example including the newly measured reference magnetism. Self-position recognition method.   The self-position recognition method according to claim 1, wherein the magnetism-position information is based on the reference magnetism measured along a predetermined path set in the environment. A device mounted on a mobile body and recognizing the self-position of the mobile body,
Storage means for preliminarily storing magnetism-position information in which a reference magnetism measured in an environment in which the moving body moves and a position where the reference magnetism is measured are associated;
A magnetic sensor that measures magnetism in the environment to measure the measured magnetism;
A position specifying means for specifying the position where the measured magnetism is measured by comparing the reference magnetism of the magnetism-position information stored in the storage means and the measured magnetism measured by the magnetic sensor;
Equipped with a,
The magnetic - location information, the self-position to X-axis, Y-axis and wherein der Rukoto what the strength and the location of the reference magnetic is associated along three axes of Z-axis respectively orthogonal Recognition device.

Images