JP2012187698A - Fresh start traveling of travelling robot, and teaching method and control method therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for a travelling robot of returning to an original route after interrupting travelling based on routing information, and then taking a fresh start travelling because the travelling robot moving from a starting point to the destination point based on the routing information having a sequence of a path and a landmark, encounters an obstacle and a part of a road under construction on the route, and becomes impossible to travel.SOLUTION: The method for performing the fresh start travelling is shown by the routinized routing information and is created by teaching. The routinized routing information is defined as the routing information in a form in which the kind and the sequence of the path or the landmark composing the routing information is fixed, but the values of position direction of beginning and terminating points of the path, and the like, are decided by the data of a sensor in execution and passage widths, and the like. The body of the travelling robot, the sensor, the computer, and the like, are always improved. It is very hard to work for modifying and testing software on site every time when improved. Manpower and time are significantly reduced because it is enough to improve the routinized routing information by the teaching.

Description

This invention is a traveling robot that moves in a place where people live or work, for example, a construction robot that is self-propelled at a construction site, an agricultural robot used for spraying agricultural chemicals and other work, a visually impaired person, an elderly person, etc. With respect to walking guide robots (guide dog robots) that guide the vehicle to the destination, and other self-running robots, when the robot fails to run based on route information, or cannot run based on route information due to obstacles or road construction It relates to re-running.

In a conventional traveling robot, route information that the robot travels from a starting point to a destination is represented by a series of paths connecting the start and end points on two-dimensional coordinates (for example, Patent Document 1) or a point series (for example, Patent Documents 2 and 3). Then, the robot moves along the route series or the point series of the route information to acquire route data. This method makes it impossible to travel based on the route information obtained by teaching when an obstacle is encountered or a road construction site is reached.

Japanese Patent No. 377083 (Japanese Patent Laid-Open No. 2002-207516) “Running robot, teaching method and control method thereof” JP-A-9-198133 JP 2000-132229 A

When a walking robot encounters an obstacle on the route or encounters a location under road construction, it must stop traveling. Therefore, a problem is how to stop traveling along the route, restart the traveling, and return to the original route.

In addition, when a walking robot travels on an uneven road surface outdoors or an indoor linoleum floor surface based on route information, position control errors occur due to wheel slipping or slipping, and the wheel rubs a curb or fence. If you come in contact with walls or pillars, you may not be able to run. In such a case, a method of interrupting traveling along the route and restarting traveling is an issue.

There are various variations of the method to redo depending on the driving environment such as the width and inclination of the passage, the shape of the intersection, and the size of the obstacle. The problem is how to express the variations on the robot system.

If you want to realize how to start over with a robot, bring the robot to the site, actually run it, and if there is a problem, you must improve it. A method of easily performing this trial and error is an issue.

The rerun traveling of the traveling robot can be expressed in several ways depending on the width of the passage and the size of the obstacle. The present invention is based on the position information of the fence or obstacle detected by the obstacle detection system and the information described in the route information when the traveling robot is too close to the fence or the entrance wall or when an obstacle is encountered. The system selects from the redo runs that have been created in advance, shows the user, asks the user whether or not to execute it, and when the user presses the redo button, runs the route to the destination. This is a method of interrupting and executing the running again, and returning to the original route when it is finished.

Further, the present invention interrupts traveling based on route information when the traveling robot comes to a location such as under construction, travels around the location under construction by guidance of an operator, etc., and travels based on the original route information. It is a method of re-running driving to return to.

Redo driving is described by route information whose elements are path, landmark, and announcement information. The route of redoing changes at that time depending on the position and size of the obstacle, the width of the road, etc. The same applies to traveling around road construction points. Therefore, there is a need for a fixed pattern of route information that is adjusted by sensor data and road environment data such as road width. Here, the fixed pattern refers to route information in which the type and order of elements constituting the route information are determined, but the values of parameters such as the position and orientation of the elements are determined by sensor data at the time of execution. Hereinafter, it is referred to as fixed path information.

By creating routine path information by teaching, the time and effort from creation to testing can be reduced. This makes it easy to improve redoing.

Advances in the body, sensors, computers, etc. of traveling robots are remarkable and constantly improved. Each time, the effort to improve the software and the labor to test in the field are great. Compared to that, the method of the present invention can be improved by changing the routine path information by teaching without changing the software, and by testing again if there is a problem, greatly reducing labor and labor. . In addition, by creating a library of routine path information, it is possible to accumulate the know-how of visually handicapped persons and gait trainers who walk independently on a daily basis.

When a robot runs on a slanted path or avoids an obstacle. Running accuracy in cm units is required. In order to improve running accuracy on uneven roads and slanted paths, the mechanism must be improved so that the springs and dampers of the vehicle body are devised to stabilize the sensor, which increases costs. If the robot has the function of redo driving, it is not necessary to perform redoing when the sensor detects an abnormality or when the user notices an abnormality. Can be suppressed.

It is a conceptual diagram which shows the structure of a traveling robot. It is a figure of the detection area of an obstacle detection system. It is a figure of the path | pass of redo driving | running | working which avoids an obstruction. An example of route information described in XML is shown. An example of route information described in XML is shown. An example of route information described in XML is shown. An example of fixed path information described in XML is shown. An example of fixed path information described in XML is shown. This is the top screen of teaching. It is a teaching root element selection screen. It is a teaching pass setting screen. It is a teaching running screen. It is a LM setting screen of teaching. It is a teaching LM confirmation screen. This is a teaching announcement setting screen. It is a flowchart which shows the whole teaching and guidance process. It is a flowchart which shows the process of an element execution subroutine. It is a flowchart which shows the process of a path | pass execution subroutine. It is a flowchart which shows the process of SP and terminal SP subroutine. It is a flowchart which shows the process of an obstruction detection subroutine. It is a flowchart which shows the process of a feedback control subroutine. It is a flowchart which shows the process of LM recognition subroutine. It is an example of a list of fixed path information names. It is a figure explaining lateral direction position shift calculation. It is a figure of a template module setting screen. It is a figure of a template module confirmation screen.

1. Outline of Embodiments The present invention relates to a method for guiding a traveling robot. When a problem occurs in which guidance cannot be continued, the traveling based on the route information is temporarily interrupted, and the traveling is restarted, that is, switched to the traveling based on the separately prepared route information. The invention is a method of returning to traveling based on the original route information when the traveling is completed.

SUMMARY OF THE INVENTION An object of the present invention is to realize route information for redoing according to various environments and conditions with a minimum amount of work and data by facilitating teaching of route information.

As shown in FIG. 1, a traveling robot according to the present invention includes a driving device composed of left and right wheels, a video camera that recognizes a sign pattern (hereinafter abbreviated as SP) and a landmark (hereinafter abbreviated as LM), an obstacle detection system, A bumper sensor that detects contact with an obstacle, an operation box that inputs a destination or route to the traveling robot, a computer that controls these devices to guide the destination, a handle for a user to hold and walk, It consists of a battery.

The operation box includes a power switch, start button, menu selection button, redo button, and joystick. Each time the user presses the start button, the robot repeats start and stop. The menu selection buttons consist of three buttons: “Previous”, “Next” and “Done”. Select a menu or list. The redo button is a button that is pressed when redoing the vehicle.

As shown in FIG. 2, the obstacle detection system divides a forward area through which the traveling robot passes from now on into a slow traveling area, a slowest traveling area, and a stop area. The slowing down region is a region where when there is an obstacle there, the type of obstacle is uttered and slowed down. The slowest traveling region is a region where the slowest traveling is performed by speaking the kind of the obstacle when there is an obstacle there. The stop area is an area where the robot stops by saying “Please come” when there is an obstacle. When the user presses the button after stopping, the vehicle starts to avoid obstacles. The width of the slow-down area is a value obtained by adding the lateral lateral movement of the robot to the lateral width of the robot. The depth is determined in consideration of the distance from the stop command received to stopping when the robot travels at a slow speed. However, unlike the case of FIG. 2, when performing a spin turn to the left and right (in-situ rotation), the slow and fastest regions are set to the left or right.

To indicate whether there is space to avoid on the left and right sides, the left (right) side turning avoidance area, left (right) side turning avoidance area, left (right) A side overturn avoidance area is provided. The small turn avoidance area includes a medium turn avoidance area and a large turn avoidance area, and the middle turn avoidance area includes a large turn avoidance area.

The computer application consists of a guidance and teaching system, an operator input system, and a route database system. The operator input system and the route database system may be installed on separate computers as shown in FIG. 1, and may be executed by connecting the traveling robot guidance and teaching system with a LAN cable.

Teaching is performed by trained operators. The operator operates the traveling robot and inputs necessary parameters according to instructions on the screen of the operator input system to create route information. When teaching is completed, it is registered in the route database. The route database system is a system that corrects route information or connects a plurality of routes created at different points of time into one route information. The guidance system is a system in which a user selects route information in a route database and guides from a starting point to a destination based on the route information.

The traveling robot is premised on teaching the route information to the traveling robot by making at least one turn along the route on which the operator should travel. The travel route of the robot is divided into a plurality of paths, a plurality of LMs, and a plurality of announcement information. A path is defined as a section that travels with one navigation. Navigation is a mode that is a unit of robot control. For example, navigation includes forward travel, reverse travel, tracking travel, spin turn, and stop. The curved route is approximated by a broken line and is represented by a plurality of forward travel paths.

A path is defined at the beginning and end. The path information can include information about the SP, information about the terminal SP, the traveling speed, the width of the passage through which the path passes, and whether or not the traveling can be performed again. However, reverse travel, spin turn and stop cannot include SP and terminal SP.

There are two types of SP, which are visual or acoustic features detected by video cameras and sensors. One is the feature that the robot travels along, such as Braille blocks, curbstones, white lines, and the like. The traveling robot corrects the positional deviation by traveling at regular intervals along the SP. The other is a feature indicating the end of a path, which is called a termination SP, and includes a mark held by a guiding person and a left or right avoidance space (a space created after obstacle avoidance is completed).

The information about the SP includes an image processing module for detecting the SP and detection parameters, an interval between the path and the SP, and a pan angle / tilt angle zoom of the video camera. The SP image processing module includes, for example, yellow braille block detection, grayscale braille block detection (detection that does not use color), road boundary edge detection, pedestrian crossing detection, and tracking running QR code detection. The image processing module at the terminal SP includes, for example, yellow warning Braille block detection, light / dark warning Braille block detection, white stop line detection, left avoidance space detection, right avoidance space detection, and tracking travel end QR code detection. However, the left (right) avoidance space is another name for the left (right) small turn avoidance area. The SP detection and termination SP detection module creates a detection parameter template for each module, and designates the template module name when teaching.

The announcement information is information related to voice guidance that is uttered when the robot arrives at a point where it is described.

The LM is used for confirmation of the current position, determination of opening / closing of a door, determination of a pedestrian signal, and the like. For confirmation of the current position, for example, a QR code is used as the LM. A QR code is pasted at the entrance and entrance of the building at the starting point or the destination, and the current position is confirmed by the robot detecting the position by image processing. The QR code is printed at the center of the paper, and a black elongated rectangle for QR code search is printed on the left and right or top and bottom and laminated. The door opening / closing determination means that a QR code is pasted on the door, and whether the door is opened or closed is determined by whether or not it can be detected by image processing. The determination of the pedestrian signal means that the color of the pedestrian signal is detected by image processing, and it is determined whether the signal is a blue signal or a red signal, the signal cannot be detected due to backlight or the like.

Information about LM: Image processing module when searching for LM, pan angle and tilde angle of video camera, zooming parameters, message uttered when QR code is recognized, retry when not recognized The number of times.

Redoing means that when the traveling robot is traveling based on the route information, if the traveling fails for some reason and the traveling based on the route information becomes impossible, the traveling is interrupted and the new route is traveled, and the cause is eliminated. Then, it means traveling where the vehicle returns to traveling based on the original route information.

Redo driving avoids obstacles, etc. automatically performed by robots, redoing that is determined by the user, for example, tracking driving to follow a person who guides during road construction, etc., and avoiding product display stands and fences at storefronts. Therefore, it is the driving | running | working which left or right.

Redo driving is expressed by routine route information. The fixed path information is one of the forms of path information, and the type and order of elements constituting the path information are determined, but the values of parameters such as the position and orientation of elements are determined at the time of execution.

The fixed path information is created by teaching the position and orientation of the starting point as (0, 0, 0) by teaching. Since the position and direction of the traveling locus of the robot at the time of guidance fluctuate along the path, the starting point of the fixed route information does not come to the point where the traveling is interrupted, but as follows. It is assumed that the travel is interrupted when the vehicle travels along the path connecting the start end (X0, Y0) and the end end (X1, Y1) and arrives at (Xc, Yc). The angle of the line segment from the beginning to the end of the path is assumed to be Dirp, and the leg of the perpendicular line dropped from the interruption point (Xc, Yc) to the path information path is assumed to be (X2, Y2). The position orientation (xt, yt, Dirt) of all the points in the fixed path information is subjected to coordinate transformation that translates to the point (X2, Y2) and rotates by Dirp during execution. Coordinate conversion from the position / direction (xt, yt, Dirt) of the fixed path information to the point (X, Y, Dir) on the coordinates of the path information is expressed by Expression (1) to Expression (6).

The redo run for obstacle avoidance is performed as follows. When the traveling robot detects an obstacle in front of the path, it speaks with the name of the obstacle, such as a roadside tree or something like a car, and decelerates. When it approaches, it pauses. The obstacle detection system detects whether there is a space for avoiding an obstacle on the left or right side of the path. If there is a space to avoid, the traveling robot asks the user whether to execute obstacle avoidance or wait until the obstacle arrives. However, if the passage width is narrow and there is no space to avoid, the suspension is continued until the obstacle arrives.

The redo run to avoid the obstacle consists of a lateral movement in front of the obstacle and a vertical movement over the obstacle. Depending on the degree of lateral movement, it will be decided whether to turn around large or small. FIG. 3 is an example of fixed path information for obstacle avoidance. The lateral movement is represented by repetition of the backward traveling path and the forward traveling path. The vertical movement is represented by a forward travel path with the left avoidance space detection as the end SP. Return to the original path at the end of the path. When the standard route information is executed at the time of guidance, the start and end of the path, the number of repetitions, and the like are not fixed, and are determined by detection data such as an obstacle detection system.

2. Route information XML expression The route information input in teaching is expressed in XML (Extensible Makeup Language).

Tag: A name tag indicating the type of data. Describe in alphanumeric characters and enclose the beginning and end of the data with </ tag> and <tag />. The data may be in Japanese (alphanumeric, kana, kanji).

Route information <PathData>: The route information includes a head <Header> and a main body <Body>. <Header> is composed of a path name <PathName> and a creation date and time <Cdate>, and the main body is composed of a series of three elements: a path <Path>, a landmark <LM>, and an announcement <Anounce>.

Creation date and time <Cdate>: Date and time when route information was created by teaching, and the date and time are necessary because SP and LM are affected by the season and time zone. Also, it is necessary to know at the time of maintenance whether or not it is old data after teaching several times.

Path <Path>: Path is navigation <Navigation>, start point <Point1>, end point <Point2>, travel speed <Velocity>, passage width left <LeftWidth>, passage width right <RightWidth>, SP information <SPInfo1>, end It consists of SP information <SPInfo2> and whether or not to start over again <RetryEnable>.

Start point <Point1> and end point <Point2>: represented by a three-dimensional vector of <X>, <Y>, <Dir>. <X> and <Y> represent the position of the robot, and <Dir> represents the orientation of the robot. The position direction of the starting point of the route is (0, 0, 0). However, the Y axis may be in the north direction, the X direction may be in the east direction, the direction may be measured clockwise, and the position may be expressed in latitude and longitude.

Path width right <LeftWidth> and path width left <RightWidth>: The width of the path on the left side and the right side of the path. One or both may be omitted. If omitted, it means that there is no width of the passage necessary for obstacle avoidance.

SP information <SPInfo1> and termination SP information <SPInfo2>: a detection module name <Module>, pan <Pan>, tilt <Tilt>, zoom <Zoom>, and a detection parameter list <Para>. Pan and tilt are the horizontal and vertical rotation angles of the video camera, and zoom is the zooming factor of the lens. The parameter list is a parameter of image processing software for detecting the SP. These values are determined by experiments prior to teaching and a template is created. If you change the height or type of the video camera, you can change the parameters of the template without changing the module.

Re-running propriety <ReteryEnable>: Indicates whether or not re-running is permitted.

LM information <LM>: LM information includes module name <module>, observation position azimuth <Point>, relative position azimuth <DevPoint>, pan <Pan>, tilt <Tilt>, zoom <Zoom>, <Found> message, It consists of a <NotFound> message, the number of trials <Try>, and a list of detection parameters <Para>. <Point> is the position and orientation of the point where the robot observes the LM, and <DevPoint> is the relative position and orientation of the LM when the observation point is the origin. <DevPoint> is represented by a three-dimensional vector of <DevX>, <DevY>, and <DevDir>. QR code information and the like are included in the <Para> list. <Message> is uttered when the LM can recognize it. <Try> is the number of times to retry when the LM cannot recognize.

Announcement <Announce>: Consists of a message label name <Message> and a waiting time <Wait>. The waiting time includes a time display such as seconds and a condition display until the user presses the button. <Message> is an identifier of voice data registered in advance.

Examples of route information described in XML are shown in FIGS. The route name is sample 1, and a path with a left and right passage width of 150 cm and 50 cm on a sidewalk with a braille block is set directly above the braille block. With the guidance Braille block set to SP, the vehicle runs straight at about 20 m at a speed of 4 km / h until a warning Braille block is detected. While traveling on this pass, it is possible to start again. Next, QR code “No. 5” in the direction of 30 degrees at 3 m ahead is recognized and recognized. When the QR code is recognized, “QR code No. 5 was found” is uttered and the recognition is terminated. If it cannot be recognized, utter it and repeat the retry up to two times. If it still does not succeed, the robot stops there, presses the redo button, and selects, for example, “wait for rescue”. If the QR code is recognized, turn 90 degrees to the right at an angular velocity of 2 km / h, say “It is the main gate of Yamanashi University”, stop for 30 seconds, and finish the run.

FIG. 7 to FIG. 8 show the fixed route information for avoiding the clockwise obstacle described in XML and the avoidance in the clockwise obstacle. The first is a reverse travel path, the start is (0, 0, 0) and the end is (20, -100, -11.5). That is, it moves backward 1m and shifts 20cm to the right. The second is a forward traveling path, the starting position is the end of the first path, the end is (40, 0, 11.5), and the position is shifted to the right by 40 cm and returns to the interrupted position. This second pass can be redone. The third path has the end position of the second path as the start end (40,0,11.5), and the left side obstacle is detected in the direction of the end (40,300,0), that is, the fault on the left side Run until you run out of things and try again and finish the routine.

When the path travel is interrupted at the time of guidance and this re-executed routine action is performed, the vehicle moves forward 40 cm to the point where the obstacle disappears, with a shift of 40 cm to the right from the interrupted point. The orientation of the robot at that time is the orientation at the time of interruption. Then start running the original path that was interrupted. However, the travel from the beginning to the end of the interrupted path is not executed. The robot advances to a point where there is a space on the left side and starts traveling on the path assuming that it is at a point shifted 40 cm to the right, so the 40 cm shift is gradually resolved and the original path is taken. When the end SP is the left avoidance space and the right avoidance space, the end of the path is the position of the point where it was detected.

If an obstacle is detected while performing the second forward travel path at the time of guidance, that is, if the obstacle cannot be avoided, the redo run is possible, so the redo run is performed again. Shift 40cm to the right. If you prepare the standard route information that shifts 20cm to the right and that that shifts 10cm to the right. By repeating the re-do routine action, it can be avoided by shifting to the right in small steps.

3. Teaching Teaching is performed in a conversational manner by the traveling robot and the operator through the display screen. Examples of display screens are shown in FIGS. When the teaching system is activated, the top screen shown in FIG. 9 is displayed and an input field for the robot name and path name is displayed. Enter characters in the input field. However, in order to distinguish the five types of fixed route information names for redo travel from the route name from the starting point to the destination, the initial letter is a special symbol, for example, a \ symbol. When the input is accepted, the creation date is displayed, and when “Next” is clicked, the route element selection screen is displayed.

The element selection screen shown in FIG. 10 displays five menus: path, landmark, announcement, and end.

When a new path is selected, a path setting screen shown in FIG. 11 is displayed. This screen displays seven icons: travel speed, left passage width, right passage width, navigation, SP detection template, end SP detection template, and whether or not to run again. Click each icon to display the numeric input field or list selection field. Click the ▽ mark to display the list one after another. The traveling speed is input in units of kilometers per hour (km / h). The navigation includes forward travel, reverse travel, left spin turn, right spin turn, tracking travel, and stop, and is selected from the list. SP detection is selected from a list of SP detection templates. The terminal SP is selected from the list of terminal SP detection templates. Whether or not redo travel is possible is determined to be rejected for a path that prohibits redo travel, such as a route crossing an intersection, and is permitted for a normal path. Point the robot in the direction of the path and click “Next” to display the teaching travel screen.

The robot displays the teaching traveling screen shown in FIG. 12 and starts traveling. This screen displays the coordinates and azimuth (X0, Y0, Dir0) of the start end of the path. Then, the SP detection screen, the terminal SP detection screen, and the current position and direction (X, Y, Dir) values are displayed momentarily. However, the SP detection screen is not displayed for a path without an SP. If the robot succeeds in SP detection, the SP recognition confirmation icon is clicked. If the terminal SP is successfully detected, the termination SP recognition confirmation icon is clicked. The robot travels to the position of the terminal SP and stops, and stores the position direction of the terminal of the path and the parameters used for detection in the path buffer. Click “Next” to return to the route element selection screen.

In the case of reverse running, the operator operates the joystick so as to run backward, and in the case of a spin turn, the operator operates the joystick to rotate right or left in place. When the end is reached, or when the desired angle of rotation is reached, the start button is pressed to inform the robot that the end of the path has been reached. The robot records path information in the path buffer.

When the landmark menu is selected on the route element selection screen, the LM setting screen of FIG. 13 is displayed. LM detection is selected from three modules: position confirmation, door opening / closing determination, and pedestrian signal determination. In either case, the operator places the traveling robot at the LM observation point.

A video camera control and LM recognition screen is displayed on the LM setting screen. The former is a screen that controls the pan, tilt, and zoom angles of the video camera. Clicking the icon Right / Left controls the pan angle, clicking Up / Down controls the tilt angle, and clicking Zoom / Tele controls the wide angle / telephoto. Is made.

The operator controls the video camera so that the image is displayed in a size that can be recognized by the LM on the imaging screen. When the traveling robot successfully recognizes the LM, an LM confirmation screen shown in FIG. 14 is displayed. This screen displays the current position direction, the pan angle, tilt angle, zoom amount, and relative position direction from the robot to the LM. The QR code recognized as this value is recorded as LM information in the path buffer. The number of retries is the number of times LM recognition is repeated when it cannot be recognized.

When an announcement is selected on the route element selection screen, the announcement setting screen of FIG. 15 is displayed. The message label name and waiting time icon are displayed on the screen. The message label name is the name of voice data uttered at the beginning of the path. Audio data is created for each path before or after teaching and is recorded with a name in the audio data library. The waiting time is selected from a list of no stop, waiting for the start button to be pressed, and a specified time stop (30 seconds, 1 minute, 5 minutes, etc.). When the above input is completed, the announcement information is recorded in the path buffer.

Teaching for redoing is performed in a path with a width that avoids obstacles. Set a route long enough to avoid obstacles, for example 10m. Place obstacles on the route. The obstacle can be anything that can be detected by the sensor, such as an automobile or a cardboard box. Put the traveling robot in the direction of the path to detect the obstacle and pause.

The fixed path information to avoid in the clockwise direction is shown in FIG. The route information name is “Right-handed obstacle avoidance”. Generate as follows. Make a path to manually retract the traveling robot diagonally to the right. Next, make a path that moves forward diagonally to the right. Next, the terminal SP creates a forward traveling path in the left avoidance space, and moves along the left side of the obstacle until no obstacle is detected. End information. When the routine information is terminated at the time of execution, the path is returned to the interrupted path. And it can return to a path | route automatically from the point far away by feedback control. However, since the passage width is the value of the passage width of the path of the original route information, it is not necessary to describe the passage width of all navigation paths used in the fixed route information.

If the obstacle is still ahead even if you start over with the standard route information, call the standard route information for the redo again, move backward and straight, move to a point where the obstacle is not ahead, and the left avoidance space Drive forward to the point where it is detected. Teaching routine route information that avoids obstacles counterclockwise in the same way.

4). Guidance guidance is a mode that uses a traveling robot teaching route information. When the user turns on the power of the traveling robot, the traveling robot speaks by voice to which route to go to the user or where the destination is. The user selects a route or a destination by operating the menu selection button. The traveling robot speaks to press the start button when the brake and clutch are in the traveling mode by voice.

When the user presses the start button, the traveling robot reads out the route information of the selected route or the destination, and executes traveling or LM recognition in order from the first path or LM. In the execution of the pass, when the SP is registered, the traveling robot detects the SP by pointing the video camera in the teaching direction. If the SP is like a curb, there is a route that is spaced to the right or left by the registered distance. If the SP is a braille block, the distance from the SP is zero and the path is in the middle of the braille block. The traveling robot performs feedback control so that the SP distance detected by the video camera matches the registered distance. When it comes near the end of the path, the SP detection is stopped, the end SP is detected, and the vehicle continues to travel to that point, and when it reaches that point, it moves to the next path or LM detection. When the destination of the last path of the route information is reached, the speech guidance is terminated.

If an obstacle enters the slow-down area in front of the path while traveling on the route, the robot decelerates to 2/3, for example, and speaks the name of the obstacle. If an obstacle enters the slow-down area, it slows down to 1/3, for example. Speak the name of the obstacle. When entering the stop area, say “Please come”.

The traveling robot speaks to the user to select with the redo button whether to move around the obstacle or wait for the obstacle to return. If the user chooses to go around the obstacle, the routine route information for avoiding the obstacle is executed, and if the obstacle is passed, the original route is returned.

When the LM is registered in the route information, when the traveling robot comes to that point, the video camera searches for the LM, performs LM recognition, and conveys the recognition result by voice. If the user presses the start button, the next pass is executed. If the LM cannot be recognized, pressing the redo button returns to the state at the end of the path before the LM recognition, and the redo routine selected by the user is executed.

When the LM is a pedestrian signal, when the pedestrian signal is red, it speaks that it is red, and when it detects that the color has changed from red to blue, it speaks to press the start button. When the user presses the start button, the traveling robot executes a path that travels across the crosswalk to the opposite bank.

When it comes to the point described in the teaching announcement, the voice data registered by voice is uttered.

5. Embodiment An embodiment in which the present invention is applied to a walking robot that guides a visually impaired person to a destination will be described below with reference to flowcharts.

4.1 Explanation of terms Guidance flag (FIG. 16): A flag (a switch on the software) that is turned on during the guidance process. Teaching processing and guidance processing are different by branching a part of the system with software guidance flags. That place is where route information input processing and output processing are performed. In the case of teaching, the operator inputs the path and landmark to be executed each time, whereas in guidance, the route information registered in the teaching process is read from the database and used. The input route information is registered in the database last in teaching, whereas it is not registered in guidance.

Interrupt flag (FIGS. 18 and 20): This flag is turned on when the obstacle detection system detects an obstacle and stops, or when the user presses a redo button and the robot stops. The flag is turned on and off in the obstacle detection subroutine.

Interrupt flag (FIG. 16, FIG. 18, FIG. 22): This flag is turned on when the interrupt flag is turned on, the conditions for redoing traveling are satisfied, and the fixed path information name is confirmed. The flag is turned on and off in the pass execution subroutine and the LM recognition subroutine.

Redo flag (FIG. 16): A flag that is turned on when the interruption flag is turned on and the process of redoing is started. This flag is already turned on when a new redo drive is performed during the redo run. The redo flag is turned on and off in the flowchart of FIG.

Slow area flag, slowest area flag, stop area flag (FIGS. 2 and 20): Flags that the obstacle detection system turns on when there is an obstacle in the area. The slow region, the slowest region, and the stop region are regions where the robot slows, slows, and stops, respectively, when there is an obstacle.

Left small turning avoidance area flag, left middle turning avoidance area flag, left large turning avoidance area flag, right small turning avoidance area flag, right middle turning avoidance area flag, right middle turning avoidance area flag, right large turning avoidance area flag (FIGS. 2 and 20): obstacle detection system Is an on flag when there are no obstacles in the area. The left (right) side small turn avoidance area is an area in contact with the left (right) side of the slow / maximum slow travel area, and is an area where the vehicle can travel if there is no obstacle. The left (right) side intermediate avoidance area is an area a little away from the slow-travel / slow-speed area, for example, 20 cm away. If there is no obstacle there, the robot can travel there. The left (right) large turning avoidance area is an area further away from the slow-travel / maximum slow-run area, for example, 40 cm away. If there is no obstacle there, the robot can travel there. The small turn avoidance area includes a medium turn avoidance area and a large turn avoidance area, and the middle turn avoidance area includes a large turn avoidance area.

Current position direction (FIGS. 18 and 21): The position (X, Y) of the local point calculated by the robot with an internal sensor (wheel encoder, gyroscope, etc.) and the direction (Dir) in which the robot body is facing Say.

Relative direction (FIGS. 18, 19, 21, 24): The relative direction refers to the difference between the direction Dir that the robot body is facing and the direction Dir0 of the path of the path.

SP interval difference (FIGS. 18, 19, 21, 24): The interval between the trajectory drawn from the start to the end and the SP is registered at the time of teaching. This is the difference between the interval value detected by the SP detection module and the registered interval value. In FIG. 24, it is represented by Dx.

SP relative azimuth (FIG. 19): a difference between the SP azimuth, Dirsp, and the azimuth of the robot body, Dirsp-Dir.

Forward gaze distance (FIGS. 21 and 24): The traveling robot is controlled not to eliminate the lateral deviation from the path of the path on the spot, but to eliminate the deviation at a certain distance in front. Say this distance. In FIG. 24, it is represented by L.

Lateral positional deviation (FIGS. 21 and 24): A value obtained by adding a positional deviation due to relative orientation at the forward gaze distance to the SP interval difference. In FIG. 24, it is represented by Dx + Ltan (Dir−Dir0).

End distance (FIGS. 18 to 20): Refers to the distance from the current position to the end of the path.

Vertical displacement (FIG. 18): A difference between the terminal distance measured by the terminal SP detection module and the distance from the current position calculated by the internal sensor to the path terminal.

End SP distance threshold ([FIG. 19]): The distance to the end SP is this value, for example, 5 cm. This is the threshold value for determining that the end of the path has been reached.

4.2 Explanation of Subroutine In the non-r-charts of FIGS. 16 to 22, the frame of the subroutine or system is represented by a double line, and its name is underlined.
Operator input subroutine (FIG. 16): Used in teaching mode (guidance flag is off). The operator inputs route information in accordance with the instructions displayed on the screen. When teaching is started, if a route name is input in accordance with the instruction on the screen shown in FIG. 9, the creation date is added to the route name and recorded in the route buffer. From now on, the operator inputs the main data of the element in accordance with the instructions on the screens shown in FIGS.

Element execution subroutine (FIGS. 16 and 17): An element is input and the element (path, landmark, announcement) is executed. The data obtained by executing the element is added to it and additionally recorded in the path buffer and output. When the obstacle does not return at the time of guidance, or when the user presses the redo button, it pauses, turns on the interruption flag, and outputs the required redo fixed path information name.

Route database search subroutine (FIG. 16): Used in guidance mode (guidance flag is on). The route name or fixed route information name is input, the route database is searched, and the corresponding route information or fixed route information is output.

Pass execution subroutine (FIGS. 17 and 18): Travels from the beginning to the end of the input path. However, if an obstacle is encountered during the execution of the path, or if the user presses the redo button, the interrupt flag is turned on and the fixed path information name is output.

LM recognition subroutine (FIG. 17, FIG. 22): When the input LM recognition module is pedestrian signal determination, the video camera is controlled to execute the pedestrian recognition subroutine. This subroutine recognizes a pedestrian signal by image processing and outputs a status code (blue signal, red signal, unrecognizable). In the present invention, a strategy is adopted in which the pedestrian signal may be crossed when it changes from red to blue. The reason is that a visually handicapped person confirms that the moving direction of the car has changed by hearing and uses the lighting time of the green light to the maximum. In the flowchart, at the time of guidance, when the pedestrian signal is blue and the red signal count is 1 or more, that is, when the signal changes from red to blue, the LM recognition subroutine is finished by speaking with a blue signal. When the pedestrian signal is a red signal, the pedestrian signal recognition is repeated until the red signal count reaches the number of trials. When the red signal remains even after the number of trials is repeated, and when neither the red nor the blue of the walking signal is recognized, the speech is made as unrecognizable. Thereafter, when the user presses the redo button, the interruption flag is turned on, the optimum redo run is selected in the routine route information name subroutine, the interruption flag and the routine route information name are output, and the process is terminated.

When the LM recognition module performs current position confirmation and door opening / closing determination, a QR code recognition subroutine is executed. This subroutine controls the video camera and recognizes code words by image processing. When it is recognized, the code word is uttered, the element is additionally recorded in the path buffer, and the process is terminated. When it cannot be recognized at the time of guidance, it repeats the recognition process as many times as the number of retries. Subsequent processing is the same as in the case of pedestrian signal determination.

Speech utterance subroutine (FIG. 17): An input message is uttered, and the subroutine is exited after a waiting time has elapsed. Announce information is additionally recorded in the route buffer and output.

SP / terminal SP detection subroutine (FIG. 18): The model SP detection module and the model terminal SP detection module are input, and the SP interval difference, relative orientation, and terminal distance are output. The SP interval difference is a difference between the SP intervals described in the template SP detection module and the detected SP interval. The SP relative direction is used only during tracking. It is the detected end SP, that is, the direction of the QR code (as viewed from the robot).

Obstacle detection subroutine (FIGS. 18 and 20): This subroutine has three functions. The first is to reduce the traveling speed of the robot. If there is an obstacle, the obstacle system detects which area is the slow speed area, the slowest speed area, or the stop area, and sets the traveling speed to slow speed, the slowest speed, or the stop as a result. The second is to select a re-do routine action to avoid obstacles. When the obstacle is in the slow-moving area, the fixed path information name selection subroutine selects whether it should be avoided on the left or right side, and if it should be avoided, whether it should be avoided by small turning or large turning. Further, when the user presses the redo button, the traveling speed is set to zero and similarly selected by the routine route information name selection subroutine. The obstacle detection system sets the current position direction to the end of the path when there is no obstacle in the left (or right) side turning avoidance area when the terminal SP is the left (or right) avoidance space detection. The third is to estimate the type of obstacle from the data of the obstacle detection system and speak. The type of obstacle is estimated from the width and height of the detected obstacle candidate three-dimensional data, for example. The input of the obstacle detection subroutine is navigation, and the output is an interrupt flag, a fixed path information name, and a traveling speed.

Feedback control subroutine (FIG. 18, FIG. 21, FIG. 24): The F04 positional deviation calculation subroutine calculates the current position direction (X, Y, Dir) by dead reckoning navigation. Then, the amount of deviation from the target trajectory is calculated when traveling in a relative azimuth (Dir-Dir0) at a forward gaze distance, for example, 2 m ahead, and set as a lateral position deviation. However, Dir0 is the direction of the orbit drawn from the start end to the end. The vehicle body control subroutine of F05 controls the left and right wheels of the robot with the lateral position deviation as a deviation. In the case of tracking traveling, the azimuth of the SP detected by image processing, that is, the QR code for tracking traveling is set as a relative azimuth, and the SP interval difference is set to zero. As a result, feedback control is performed assuming that there is a QR code code at the forward gaze distance.

Standard route information name selection subroutine (FIGS. 20, 22, and 23): (Subroutine for outputting a corresponding standard route information name with the left / right (small / medium / large) turn avoidance area flag and the redo button as inputs) However, when the redo button is on, the user is made to select from the list of fixed path information names shown in FIG.

4.3 Description of Flowchart Teaching processing will be described with reference to the flowchart of FIG. A sighted person who teaches, for example, a walking trainer, is called an operator. The operator designs what kind of path and landmark the route is to be decomposed, and places the traveling robot in the direction of the first path from the starting point.

Teaching starts from A01 in FIG. In A02, the guidance flag is turned off, and in A03, the operator inputs a route name. The operator branches by the guidance flag A04, and the operator inputs an element name and information on the element from the operator input system A05. The operator input system is a system for inputting element data in accordance with instructions on the PC screen shown in FIGS. Proceed to the element execution subroutine of A07. Details of this subroutine are shown in FIG. The input of the subroutine is an element, and the output is a path buffer, an interrupt flag, and a fixed path information name. The element data obtained by teaching is recorded in the path buffer in the order of acquisition. The interrupt flag is always off during teaching. Branch at A08 and return to A04. This loop repeats until the end of the path is reached and the operator empties the element. Through A09, the route buffer is registered in the route database at 10 and the teaching process is terminated at A11.

The guidance process will be described. Starting at A13 in FIG. In A14, the guidance flag is turned on. In A15, the user speaks to select a route name. In A16, the route database is searched to extract route information corresponding to the route name. Since the guide flag of A04 is on, an element is extracted from the route information at A17. If there is an element in A06, the process proceeds to the element execution subroutine to execute the element. In normal driving, the output interruption flag is off, and the process returns to A04. When the route information reaches the end, the element disappears in A06, and the process proceeds to A12 through A09. Since the driving is not redone, the utterance is reached in A18 and the guidance procedure ends in A19.

If the interruption flag is turned on while the element is being executed in the element execution subroutine of A07 in the guidance process, the process proceeds to A20 via A08. If the redo flag is off, the process proceeds to A21, and this flag is turned on. The path information interrupted at A22 and the path information are saved in the interrupt buffer, and the process proceeds to A23. However, if the interrupted element is LM, the previous path information is saved in the path buffer.

A23 is a route information retrieval subroutine. This subroutine searches the route database with the re-input fixed route information name and outputs the fixed route information. In A24, the robot moves in parallel so that the starting point of the standard route information comes from the point of interruption to the point of the foot of the perpendicular line dropped to the path. Then, the entire fixed path information is rotated in the direction of the interrupted path. Next, it progresses to A04 and the guidance based on the fixed path information is performed from A17. When it reaches the end of this fixed path information, it branches to No at A06, proceeds to A12 via A09, and the redo flag is on, so the redo flag is turned off at A25. Then, the interrupted route information and the path are returned from the interrupt buffer, and the traveling by the original route information is continued.

The redo flag is on while the redo action is being executed. In this case, when the element execution subroutine of A07 turns on the interruption flag and redoes and outputs the fixed path information name, the process proceeds to A23 via A20. That is, even if a new redo run is executed during the redo run, the pass when it is finished and returned is the pass when it was first interrupted.

The element execution subroutine of FIG. 17 is divided into three subroutines depending on the type of element. If the element is a path, the path execution subroutine of B03 is executed. This subroutine ends after traveling on the track from the start to the end based on the path information. However, when an obstacle is encountered while traveling and the redo button is pressed, the suspension flag is turned on, the traveling is suspended, the fixed path information name is output, and the processing is terminated. If the element is LM, the B04 LM recognition subroutine is executed. This subroutine recognizes the LM, and when it is recognized, utters the recognition result and ends. When it cannot be recognized, the recognition is repeated for the number of retries. If it still cannot be recognized, the user presses the redo button to select the fixed path information name, turns on the interruption flag, and ends. If the element is an announcement, a speech utterance subroutine is executed, a message is uttered, and the process ends when the waiting time has elapsed. In any subroutine, element information obtained by the execution is additionally recorded in the path buffer.

The path execution subroutine of FIG. 18 is first divided into three processing groups by the navigation of C02. The first group consists of forward travel, back travel, and tracking travel. The second group consists of a left spin turn and a right spin turn, and the third group consists of a stop.

The first group creates a target trajectory connecting the beginning and end of the path at C03. In C04, the interruption flag is turned off, and the process proceeds to the SP / terminal SP detection subroutine of C05. This subroutine activates the SP detection module and the termination SP detection module. The SP detection module recognizes the SP and detects an interval between the SP and the path. The difference between the interval and that registered by teaching is calculated as the SP interval difference. On the other hand, the terminal SP detection module detects the distance to the terminal SP, and sets the distance as the terminal distance. This subroutine receives the SP detection module and the termination SP detection module as inputs, outputs the SP interval difference and the termination distance, and proceeds to the obstacle detection subroutine of C06.

In the obstacle detection subroutine of C06, when the obstacle detection system detects an obstacle, the traveling speed is set to the slow speed, the slowest speed, and the stop (travel speed = 0) according to the position. Even if there are no obstacles, it stops when the user presses the redo button. When the redo button is pressed, the interrupt flag is turned on, a fixed path information name is output, and the subroutine is terminated.

When the interrupt flag is on in C07, the presence / absence of fixed route information is checked, and if there is a next road width to be avoided, if the user presses the redo button again, the interruption flag is turned on, Output the fixed path information name and end the path execution subroutine. If the redo button is not pressed again, the process returns to the obstacle detection subroutine of C06, and the obstacle detection is repeated until the obstacle retreats. When there is no fixed route information name and when the road width is narrow, the processing returns to the C06 obstacle detection subroutine in the same manner as an utterance that obstacles cannot be avoided.

When the interrupt flag is off, the C08 feedback control subroutine is executed. When the running speed is zero, do nothing and finish. The trajectory connecting the beginning and end of the path is set as a target value, and the lateral position deviation at the forward gaze distance is used as a deviation to perform feedback control at the traveling speed. If the distance to the end is larger than the end threshold in C09, the process returns to C05. If the distance is smaller, the vertical position deviation of the current position is corrected in C10, and the element is additionally recorded in the path buffer in C20, and the path execution subroutine is completed. .

If the navigation is a left or right spin turn, the left and right wheels rotate on the spot at angles designated in opposite directions at C17, and the subroutine is terminated. When an obstacle is detected during rotation, the interrupt flag is turned on in the obstacle detection subroutine at C18, and the process proceeds to C13 to determine whether to press the redo button, and the path execution subroutine is terminated.

When the navigation is stopped, the travel is stopped, and when the start button is pressed, the pass execution subroutine is terminated and the next element is executed.

The SP / end SP subroutine of FIG. 19 ends if the end distance is shorter than the end SP threshold. Otherwise, the SP module is activated, and when SP cannot be detected, the process proceeds to D09. If detected, the process proceeds to D05. When the navigation is tracking traveling, SP is a QR code for tracking traveling, the SP relative direction is detected in D07, and the value is relative direction. If it is not tracking running, the relative azimuth and SP interval difference are calculated in D08. Then, the process proceeds to D09.

The terminal SP module is activated at D09. The terminal SP of the tracking traveling is a QR code for ending the tracking traveling. At the point where the QR code changes from tracking to end, the end distance is set to zero and the path ends. When the navigation is not tracking traveling, the distance to the terminal SP is the terminal distance, and when the terminal SP cannot be detected, the terminal distance is the distance from the local point to the path terminal, and the subroutine is terminated.

The obstacle detection subroutine of FIG. 20 turns off the interrupt flag at E02. The obstacle data is received from the obstacle detection system. The robot traveling direction area is divided into a slowing area, a slowest traveling area, and a stopping area. If there is an obstacle in the area, the area flag is turned on. A small turn avoidance area, a middle turn avoidance area, and a large turn avoidance area are provided on the left and right sides of the slow / stop area. If there is no obstacle in each avoidance area, the avoidance area flag is turned on.

In the re-executed fixed route information, when returning to the original route while avoiding the obstacle, the forward traveling path is used, and the terminal SP is set as the left or right avoidance space detection. When the left edge is set to the right-handed corner avoidance area flag, the terminal distance is set to zero. That is, it is assumed that the end of a path that avoids an obstacle is reached.

When the robot moves forward and the obstacle reaches the stop area, the obstacle's front field of view is blocked, and it is impossible to know whether there is an avoidance space on the left or right side of the obstacle. Therefore, at the point where the obstacle enters the slow-moving area, whether to avoid a large turn or a small turn is selected in the fixed route information name selection subroutine of E15. However, the interrupt flag is actually turned on at the point E08 where the stop area flag is turned on. Regardless of the presence or absence of an obstacle, when the user presses the redo button, the traveling speed is set to zero and the fixed route information name is selected in E15. When the slow speed, maximum slow speed, and stop area flags are on from E07 to E11, the respective traveling speeds are set, the obstacle name is uttered or a warning sound is sounded, and the subroutine is terminated.

The feedback control subroutine of FIG. 21 sets the SP interval difference to zero during tracking traveling. In the F04 positional deviation calculation subroutine, the lateral positional deviation at the forward gaze distance is calculated. Next, the feedback control is performed by setting the path trajectory to the target value by making the speed control and the lateral position deviation a deviation in the vehicle body control of F05.

4.4 Model Module Description The SP and end SP detection module are created by the system developer. For example, two module names are input on the screen shown in FIG. 25, and the SP and path interval are input as parameters. The interval identifies whether the SP is on the right or left of the path by the sign of the value. When the next button is clicked, the screen shown in FIG. 26 is displayed. The robot travels the test distance and activates the SP and end SP detection module. The screen appears when you operate the video camera with the video camera control icon. When the “success” icon is clicked when the detection is successful, the pan / tilt / zoom angle at that time and the detection threshold values of Para1 to Para3 are recorded as values of the template module. When the detection fails, the video camera control and the detection threshold are changed, and “Return” is clicked and repeated until the detection is successful.

The lines from <SPInfo1> in FIG. 4 to </ SPInfo2> in FIG. 5 are XML representations of the guidance Braille block detection and warning Braille block templates input in FIGS.

Claims (3)

Expressing the route information of the traveling robot as a series with the path and landmark as elements, from the path connecting the start and end of the path, the forward, backward, spin turn, tracking after the mark, navigation consisting of stop, and from the trajectory The sign pattern information for correcting the lateral deviation of the trajectory, the information of the terminal sign pattern for correcting the positional deviation in the vertical direction of the trajectory, and the traveling speed, the width of the passage, etc. A traveling robot that creates route information by moving a traveling robot along a route. When traveling based on route information from the starting point to the destination point, when you are blocked from reaching an obstacle or come into contact with an entrance or a fence, the information detected by the obstacle detection system is used to restart the vehicle. Automatically or when the user chooses, or when he / she comes to a construction site on the road, he / she chooses a re-run that consists of a tracking run according to the guidance of the construction worker, and runs based on the route information to the destination point. The traveling robot according to claim 1, wherein the traveling robot is interrupted, re-executed, and the traveling robot returns to traveling based on the suspended route information when the problem disappears. The route information for redo driving is described in the standard route information, that is, the type and order of the constituent elements, but the position and orientation values such as the start and end points of the path are not determined, and the route information is determined at the time of execution. The traveling robot according to claim 2, wherein the fixed path information is created by teaching.

Images