JP2006252348A - Mobile robot - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce influence due to the shift of an obstacle detection position that accompanies the movement, preventing an obstacle from being misdetected. <P>SOLUTION: This mobile robot 1 is provided with a storage part 15 for storing an environment map 73 corresponding to a prescribed environment, an own position detecting part 9 for recognizing a position on the environment map corresponding to own position, an obstacle detecting part 11 for scanning the environment to detect a relative position of a measured object, a voting part 91 for adding a vote value to a position on the environment map, corresponding to the detected relative position of the measured object, a vote value calculating part 97 for weighting the vote value by the voting part 91, based on the the detected relative position of the measured object; and an object determination part 93 for determining the presence of an object in the position on the environment map, when the sum of the added voted values become a prescribed threshold or larger. The voted value calculating part 97 calculates the voted value by the weighting in accordance with the decreasing function, such as inverse proportionality of reducing the weight, along with the increase in the distance up to the detected measuring object. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a mobile robot, and more particularly to improvement of object detection capability in a mobile robot.

  In recent years, mobile robots that move along a predetermined route while detecting an obstacle have been provided. As this mobile robot, an obstacle detection sensor that irradiates various exploration signals such as laser beams, visible rays, ultrasonic waves, infrared rays, etc. to the traveling direction and surroundings of the robot and receives a reflected regression signal from the object as a detection signal And detecting an obstacle near the route by the input of this sensor is known.

With regard to such a mobile robot, Patent Document 1 specifically detects an existing object as an obstacle by setting in real time a detection area that excludes the existing object using an existing object map in which the existing object is set in advance. There is described an automatic guided vehicle that prevents an emergency stop when an obstacle other than an existing object is detected by scanning with a laser beam.
Japanese Patent Laid-Open No. 9-6433 (FIG. 2)

  The above-described automatic guided vehicle in the conventional document includes a laser ranging sensor as an obstacle detection sensor, receives a reflected laser beam reflected from the object surface, and calculates the relative position of the object from the distance and direction. It has become.

  However, the automatic guided vehicle moving along the route may be slightly or momentarily misaligned due to rattling, blurring, or the like as it travels depending on road conditions. Such an instantaneous or minute deviation is difficult to detect by the automatic guided vehicle itself. Moreover, in the obstacle detection sensor that scans the periphery of the automatic guided vehicle in order to detect the obstacle, the self-direction recognized by the automatic guided vehicle itself even if the automatic guided vehicle is slightly shifted ( The angle between the (posture) and the actual orientation (posture) is greatly shifted, and the measured value of the sensor changes greatly.

  Therefore, the automatic guided vehicle in the conventional document may detect an object at a relative position different from the actual position of the obstacle when the direction is momentarily or slightly deviated as the vehicle travels, which may cause erroneous detection. . In particular, as the relative position of the detected object is farther from the automatic guided vehicle and the relative position of the detected object deviates from the front of the sensor, the influence of the deviation of the direction of the transport vehicle increases and the risk of false detection increases. Become.

  In addition, since the automatic guided vehicle of the conventional document is for detecting an obstacle existing in the detection area by setting a detection area excluding the existing object, an abnormality occurring in the existing object, for example, disappearance of the existing object or Damage could not be detected.

  Therefore, an object of the present invention is to provide a mobile robot that can reduce the influence of the displacement of the obstacle detection position accompanying the movement to prevent the erroneous detection of the obstacle and can also detect an abnormality occurring in the existing object.

  The present invention relates to a robot that moves in a predetermined environment by traveling means. The mobile robot stores an environment map corresponding to the predetermined environment, and the environment map corresponding to its own position. A position recognition unit for recognizing the upper position; a detection unit for scanning the environment to detect a relative position of the measurement object; and the environment corresponding to the relative position at which the measurement object is detected by the detection unit A voting unit that adds a voting value to a position on a map, a voting value calculating unit that weights a voting value by the voting unit based on a relative position of the object detected by the detecting unit, and the added The object determination unit that determines that an object is present at a corresponding position on the environment map when the accumulated voting value is equal to or greater than a predetermined threshold, and the voting value calculation unit is provided in the detection unit. Distance to the measured object It calculates a voting value by weighting according to the decreasing function to reduce the weight more increases. The position on the environmental map may be expressed by an area (grid) obtained by dividing the map into a plurality of parts.

  According to the above configuration, when the relative position of the object to be measured is detected, the vote value is added to the position on the environment map corresponding to the detected relative position, and the vote value is equal to or greater than the threshold value. It is determined that there is an object. And in particular, since the voting value is given a weight according to a decreasing function that decreases the weight as the distance to the object to be measured increases, even if the detection position shifts as the robot moves, By lowering the weight for a distant measurement point where the influence of the angle deviation becomes large, the influence of the deviation of the detection position can be reduced to prevent erroneous detection.

  The voting value calculation unit may calculate the voting value by weighting that is inversely proportional to the distance to the object to be measured detected by the detection unit, and accordingly, an appropriate weighting that makes the voting value smaller at a distant measurement point. Can do.

  The voting value calculating unit may further calculate the voting value by weighting according to a decreasing function that decreases the weight as the distance between the relative positions of the adjacent objects to be measured detected by the detecting unit increases. More specifically, the voting value calculation unit decreases the weight as the distance between the relative positions of the measured objects measured before and after the detection unit scans at a constant time interval increases. The vote value may be calculated by weighting according to the function. As a result, even if the detection position is displaced due to the movement of the robot, the influence of the displacement of the detection position is reduced by reducing the weight for the measurement points that are adjacent to each other so that the influence of the angle displacement is large. It can reduce and prevent false detection.

  Further, the voting value calculation unit may calculate the voting value by weighting inversely proportional to the distance between the relative positions of adjacent objects to be measured. Appropriate weighting is possible.

  In the present invention, the storage unit may store existing object information having position information of an existing object in the predetermined environment, and is different from a position on the environment map corresponding to the existing object information. An abnormality determination unit may be provided that determines that there is an abnormality when the object determination unit determines that an object exists at a position. Accordingly, it is possible to prevent an existing object from being detected as an obstacle and detect only an abnormal object.

  In the present invention, the abnormality determination unit may include a position on the environment map corresponding to the relative position of the object to be measured detected by the detection unit, and the environment map recognized by the position recognition unit. It may be determined that there is an abnormality when a position on the environment map corresponding to the existing object information exists on a straight line connecting the self position. Thereby, it can be determined that an existing object that should exist can not be detected, and an abnormality such as disappearance or breakage occurring in the existing object can be detected.

  In the present invention, the search range of the position corresponding to the existing object information on the straight line may be set from the self position to a position a predetermined distance before the relative position. Can be prevented.

  In the present invention, the abnormality determination unit may determine that there is no abnormality when the position corresponding to the existing object information continues to the end point of the search range on the straight line. Can be prevented.

  Further, in the present invention, in the mode for generating the existing object information, the existing object registration for registering the existing object information of the existing object in the storage unit as an existing object determined to be present by the object determining unit And the threshold for determination by the object determination unit is set larger when detecting an object for abnormality determination by the abnormality determination unit than when detecting an existing object. Good. Thereby, the object detection function for detecting an abnormality can be utilized for acquiring existing object information, and the existing object information can be easily obtained and stored in the storage unit. In addition, by setting the threshold value at the time of abnormality detection larger than the threshold value at the time of acquiring existing object information, it is possible to appropriately adjust the sensitivity of abnormality detection and reduce erroneous detection.

  In the present invention, the storage unit stores a voting table that divides the environmental map into a plurality of regions and stores voting values for each region, and the voting unit detects the object detected by the detection unit. The voting value is added to the area corresponding to the relative position of the measurement object and stored in the voting table, and the object determination unit corresponds to the area when the voting value in the voting table becomes a predetermined threshold value or more. A combined region extraction unit that extracts regions adjacent to each other on the environment map as a combined region, the object determination unit including: In addition, it may be determined whether or not an object exists in the combined area based on a total sum of voting values given to each area constituting the combined area extracted by the combined area extraction unit. This makes it possible to detect the presence of a moving object that moves to an adjacent area before the vote value of one area exceeds the threshold. For example, the combined area extraction unit may extract a circumscribed rectangle surrounding the adjacent areas as the combined area.

  In the present invention, the object determination unit determines that there is no object in the corresponding area when the number of positions where the vote value is equal to or greater than a threshold value is within a predetermined number within a predetermined range. Thus, even if the vote value is equal to or greater than the threshold value, it can be determined that there is no object at an isolated position, and erroneous detection of the object can be prevented.

  Another aspect of the present invention is an object detection apparatus, which detects a relative position of an object to be measured by scanning a predetermined scanning range and a storage unit that stores an environment map corresponding to the predetermined environment. A voting unit for adding a voting value to a position on the environmental map corresponding to a relative position of the object to be measured detected by the detecting unit, and a object to be measured detected by the detecting unit. A voting value calculation unit that weights the voting value by the voting unit based on a relative position, and an object at the corresponding position on the environment map when the accumulated voting value is greater than or equal to a predetermined threshold value. The voting value calculating unit that determines that the voting value is calculated by the weighting according to a decreasing function that decreases the weight as the distance from the object detected by the detecting unit increases. Is calculated. Also in this aspect, the above-described effects of the present invention can be obtained. The object detection device may be mounted on a mobile robot, and thereby the above-described mobile robot of the present invention is configured. However, this aspect is not limited to such a configuration. The object detection device may be mounted on an arbitrary moving object, and may be fixed to an appropriate place such as a wall surface. The object detection device may be an obstacle detection device. Various specific configurations described above may also be applied to this aspect.

  In addition, the present invention is not limited to the above-described mobile robot and obstacle detection device. The present invention may be expressed and specified in the form of a method, program or system. For example, another aspect of the present invention is an object detection method or abnormality determination method that performs the operation of the above configuration. Another aspect of the present invention is a method for controlling a mobile robot. Another aspect of the present invention is a program that causes a computer to execute such a method. Another aspect of the present invention is a system in which a mobile robot and a monitoring center are connected by communication.

  As described above, according to the present invention, it is possible to reduce the influence of the displacement of the obstacle detection position due to movement or the like and prevent erroneous detection of the obstacle, and also to detect an abnormality occurring in the existing object. It becomes possible to do.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. The mobile robot according to the present embodiment travels along a predetermined route in a predetermined environment serving as a monitoring area, detects an object that appears or disappears in the monitoring area, and detects an abnormality. When the mobile robot detects an abnormality, it sends an abnormality signal to a remote monitoring center. When the monitoring center receives the abnormality signal, the monitoring center checks the image by the imaging means mounted on the mobile robot, or remotely operates the mobile robot to deal with the abnormality.

  The configuration of the mobile robot is shown in FIG. The mobile robot 1 includes a moving means 3, a movement control unit 5, a guide detection unit 7, a self-position detection unit 9, an obstacle detection unit 11, an obstacle determination unit 13, a storage unit 15, a communication unit 17, an imaging unit 19, and the like. It is comprised from the control part 21 which controls each part, and the power supply part 23 which supplies electric power to each part. Each part will be described below.

  The moving means 3 includes motors 35 and 37 that independently drive the right wheel 31, the left wheel 33, and the left and right wheels. As shown in FIG. 2B, the mobile robot 1 has four wheels, two of which are drive wheels, and correspond to the right wheel 31 and the left wheel 33 in FIG. These driving wheels rotate and the mobile robot 1 travels. The straight traveling speed and the turning traveling speed are controlled by the rotational speeds of the left and right wheels, and the turning direction is also controlled. The rotational speeds of the left and right wheels are independently controlled by the movement control unit 5. Instead of controlling the left and right wheels independently as described above, a method of controlling the turning angle by controlling the steering angle may be adopted. Further, instead of wheel driving, a method of independently controlling the left and right crawlers may be employed.

  The guide detection unit 7 detects guide means on the movement path. The guide means is shown in FIG. As shown in the figure, a white line tape 101 as a guide means is fixedly provided over the entire length of the path on the movement path that the preset mobile robot circulates. In addition to the white line tape 101, an indication marker 103 as a point indication means is fixedly provided at a predetermined point in the route. The indication marker 103 is a pair of white and square marks, and is provided at the boundary of the section set on the movement route. In the present embodiment, for example, a point that moves from a straight line to a curve and the opposite point are set as the boundary of the section, and an instruction marker 103 is provided.

  The guide detection unit 7 includes a white line detection camera 41 and an image processing unit 43. The white line detection camera 41 is installed on the bottom surface of the mobile robot 1 so that the road surface can be photographed. The image processing unit 43 is realized by a computer of the mobile robot 1 and has a white line tape 101 to be used to guide the path of the mobile robot 1 from a photographed image of the white line detection camera 41 by processing such as edge extraction and Hough conversion, and an indication marker. 103 is detected and output to the control unit 21.

  In addition, the guide detection part 7 is not limited to said structure. The guide detection unit 7 may be configured by a magnetic sensor, an electromagnetic induction sensor, or the like. In this case, the guide detection unit 7 may be configured to detect a magnetic guide or an electromagnetic induction guide as guide means installed in the movement path. It is preferable that the guide means and the guide detection unit can be selected depending on the installation environment.

  The movement control unit 5 is a means for controlling the drive motors 35 and 37. The movement control unit 5 controls the drive motors 35 and 37 to move along the white line tape 101 by, for example, well-known PID control according to the detection output of the white line tape 101 by the guide detection unit 7. Further, the movement control unit 5 controls the movement speed based on the route information set in advance in response to the detection of the travel section by the self-position detection unit 9. The moving speed is preset for each section and stored as part of the route information.

  The self-position detection unit 9 includes resolvers 51 and 53 and a position calculation unit 55, and functions as a position recognition unit. The resolvers 51 and 53 are installed in the motors 35 and 37, respectively, and detect the absolute positions of the motor rotation shafts of the motors 35 and 37, respectively. The resolvers 51 and 53 are one form of the rotation amount detection unit. The position calculation unit 55 is realized by a computer of the mobile robot 1 and calculates the rotation amounts of the left and right wheels 31 and 33 from the rotation amount of the motor rotation shaft obtained from the resolver output. The position (Xr, Yr) and posture (θr) are calculated. In this process, the travel distance and the angle change are obtained from the wheel rotation amount, and the position and posture at each time point may be captured from these information, and the angle change is determined by the difference in the wheel rotation amount and the driving wheel distance ( Tread). Such position detection is a technique generally known as dead reckoning (autonomous navigation).

  Further, the position calculation unit 55 counts the number of detections of the indication marker 103 in accordance with the detection output of the indication marker 103 by the guide detection unit 7, detects the current travel section based on the route information from the detection number, and moves the movement route Recognize the start and end points, and correct the self-position.

  The obstacle detection unit 11 is means for detecting an object in front of the mobile robot 1 and corresponds to the detection unit of the present invention. The obstacle detection unit 11 includes a laser sensor 61 installed toward the front of the mobile robot 1 and receives reflected light when the laser light emitted from the laser oscillator is reflected by an object on the optical path. The obstacle detection unit 11 controls the irradiation direction of the laser light emitted from the laser oscillator by the scanning mirror and the means for rotationally driving the scanning mirror, so that a predetermined range including the front of the mobile robot 1 is set to, for example, 30 msec. Spatial scanning is performed at a period of.

  The obstacle detection unit 11 then scans the distance between the obstacle detection unit 11 calculated from the time from laser light irradiation to reflected light detection and the object (measurement point) reflecting the laser light, and rotationally driven scanning. Based on the angle of the mirror, the relative position of the object reflecting the laser beam, that is, the measurement point reflecting the laser beam is calculated. The relative position is the position of the measurement point with respect to the mobile robot 1.

  The obstacle detection unit 11 may be configured with a sensor other than the laser sensor. For example, the obstacle detection unit 11 may be configured by an infrared type sensor or a millimeter wave radar type sensor.

  The storage unit 15 stores information used for various processes in the mobile robot 1. The information stored in the storage unit 15 indicates route information 71 indicating information on a moving route, an environment map 73 indicating a monitoring area in a two-dimensional coordinate system, and a position of an object such as an existing object measured at a normal time. The existing object information 75, a voting table 77 for voting the measurement result of the obstacle detection unit 11, and current position information 79 including the position and posture of the mobile robot calculated by the position calculation unit are included.

  As the route information 71, corresponding to the number of each section (section from one indication marker to the next indication marker) in the movement route, the traveling speed of each section, the section distance measured in advance, the direction between the start point and end point of the section The angle difference (angle difference) is stored.

  The environmental map 73 is map information of a monitoring area, and is partitioned by a grid for each predetermined distance, and an identification number is given to each grid. In this embodiment, an example in which an environmental map is partitioned by a 5 cm × 5 cm grid will be described. For example, when dividing every 5 cm × 5 cm, the environmental map of the 500 m square monitoring area is divided into 100 million grids.

  The existing object information 75 is information indicating the position of an object such as an existing object for each grid. In the present embodiment, the existing object information 75 is information that is developed on the environment map 73 and recognized as an attribute of the grid of the environment map 73 and is a part of the environment map 73. That is, each grid of the environment map 73 has attribute information indicating whether or not an existing object exists therein, and this attribute information is used as existing object information.

  FIG. 3 is a schematic diagram showing the environment map 73. In FIG. 3, only the peripheral part of the mobile robot 1 is shown. The hatched grid in the figure is a grid in which the position of an object is indicated by existing object information, that is, an existing object grid.

  The existing object information 75 is generated by the mobile robot 1 itself through measurement performed by the mobile robot 1 in a normal state. That is, the mobile robot 1 travels on the travel route for registering the existing object information 75, and the position of the object detected by this travel is stored in the storage unit 15 as the existing object information 75.

  The voting table 77 is a table in which the measurement result of the obstacle detection unit 11 is voted for each grid. Each grid of the voting table 77 corresponds to each grid of the environment map 73. The vote value of each grid corresponds to the number of times the obstacle detection unit 11 has obtained the measurement points in the corresponding grid.

  The communication unit 17 is a wireless communication unit that transmits and receives signals to and from a remote monitoring center. When the obstacle determination unit 13 detects an abnormality, the communication unit 17 outputs an abnormality signal wirelessly to a remote monitoring center. The communication unit 17 transmits an image captured by the imaging unit 19 to a remote monitoring center, and inputs a control command received from the monitoring center to the control unit 21.

  The imaging unit 19 is an imaging unit that is mounted on the mobile robot 1 and images the surroundings of the robot. For example, as shown in FIG. 2B, the imaging unit 19 is housed in a hexagonal prism housing at the top of the robot. In the housing, six image pickup units are accommodated in six directions, so that the entire visual field in the horizontal direction is covered. The image captured by the imaging unit 19 is buffered in the storage unit 15 and transmitted from the communication unit 17 to the monitoring center.

  The obstacle determination unit 13 includes a voting unit 91, an object determination unit 93, and an abnormality determination unit 95, and the voting unit 91 includes a voting value calculation unit 97. The obstacle determination unit 13 determines whether there is an abnormality based on the output of the obstacle detection unit 11. When an abnormality is determined, an abnormality signal is output from the communication unit 17, and the movement control unit 5 performs predetermined processes such as movement stop and deceleration, obstacle avoidance and tracking.

  When the obstacle detection unit 11 measures the measurement point, the voting unit 91 determines a grid corresponding to the relative position of the measured measurement point based on the position and orientation of the mobile robot 1 from the environment map 73, and the voting table The vote value is added to the corresponding grid of 77.

  The object determination unit 93 determines the presence or absence of an object in each grid based on the vote value for each grid in the voting table. And the object determination part 93 determines with an object existing in the grid whose vote value became more than a predetermined threshold value. In addition to functioning when determining an abnormality, the object determination unit 93 also functions when the mobile robot 1 travels along a route to detect an existing object and register existing object information 75 prior to determining the abnormality.

  The abnormality determination unit 95 functions together with the object determination unit 93 to determine abnormality. At this time, the abnormality determination unit 95 can determine the presence or absence of an intruder other than an existing object, and functions as an intruder abnormality determination unit. Furthermore, the abnormality determination unit 95 can determine whether there is an abnormality based on the position and posture of the mobile robot 1 and the existing object information 75 that should be measured. At this time, the abnormality determination unit 95 can determine the disappearance of the existing object and functions as a loss abnormality determination unit.

  The voting value calculation unit 97 is a unit that weights the voting value by the voting unit 91 based on the detected measurement point of the object to be measured, more specifically, based on the relative positional relationship between the mobile robot 1 and the measurement point. . The vote value calculation unit 97 calculates a weighted vote value, and the vote value is added by the vote unit 91.

  The voting value calculation unit 97 performs weighting processing according to a decreasing function that decreases the weight as the distance from the laser sensor 61 to the measurement point increases. As shown in FIG. 13A, the decrease function is specifically inversely proportional. That is, the vote value is calculated so as to be inversely proportional to the distance from the laser sensor 61 to the measurement point.

  The meaning of this weight is as follows. If the accuracy of the self-position detection unit 11 is not sufficient, there is a possibility that the position of the measurement point may be shifted due to a slight fluctuation of the posture θr of the robot. The effect of this phenomenon is so great that the measurement point is located farther from the mobile robot 1 (see FIG. 4). Therefore, by giving a weight inversely proportional to the distance to the measurement point, it is possible to reduce the influence of the deviation of the measurement point due to the shake of the posture θr of the mobile robot 1.

  In the present embodiment, the voting value calculation unit 97 calculates the voting value obtained by performing the following weighting in addition to the weighting according to the distance. That is, the voting value calculation unit 97 performs a weighting process according to a decreasing function that decreases the weighting as the distance between adjacent measurement points increases. As shown in FIG. 13 (b), the decrease function is also specifically inversely proportional here. In this case, the distance between the current measurement point and the previously detected measurement point is obtained. The vote value is calculated so as to be inversely proportional to the distance between these measurement points. As a result, a vote value that is inversely proportional to the distance between the measurement points obtained before and after the laser sensor 61 scans at a constant time interval is obtained.

  The meaning of this weight is as follows. Even if the distance of the measurement point itself is long, the measurement point facing the laser sensor 61 is not easily affected by the shake of the posture θr of the mobile robot 1. On the other hand, measurement points that do not face each other, that is, measurement points that have a larger angle from the facing direction of the laser sensor 61, are greatly affected by the shake of the posture θr of the mobile robot 1 (see FIG. 5). If the distance is the same, the closer to the front of the laser sensor, the shorter the distance between the measurement points and the less the influence of the posture blur. Thus, as described above, weighting according to the distance between the measurement points can be applied to reduce the influence of the measurement point shift due to posture blurring.

  In the present embodiment, both of the two weightings described above are performed. For example, a reference value (for example, 1) for the vote value is set. Then, a table (or map) of distances and weighting coefficients is obtained in the storage unit 15 so that the vote value of FIG. Further, a table (or map) of distances between measurement points and weighting coefficients that can obtain the vote value of FIG. 13B is stored in the storage unit 15. The weighting coefficient corresponding to the distance from the laser sensor 61 to the measurement point is read from the former table, and the weighting coefficient corresponding to the distance between the measurement points (previous and current) is read from the latter table. Then, the voting value is calculated by multiplying the reference value by two weighting factors. This vote value is voted on the vote table by the voting unit 91.

  Here, both of the two weightings are performed, but one of the weightings may be performed within the scope of the present invention.

  Returning to the description of FIG. 1, the control unit 21 is a unit that controls the configuration of each unit of the mobile robot 1 described above, and is configured by a computer including a CPU and the like. What can be computer-processed by each part structure of FIG. 1 may be implement | achieved by the same computer. For example, the image processing unit 43 of the guide detection unit 7 and the position calculation unit 55 of the self-position detection unit 9 are realized by the above computer, and the functions of a part of the obstacle detection unit 11 and the obstacle determination unit 13 are the same. It may be realized with a computer. The same applies to the movement control unit 5. For example, an upper command value related to motor control may be calculated and transmitted to the motor drive circuit. Furthermore, the storage unit 15 may be realized by a memory of the computer and an external storage device.

  Next, the operation of the mobile robot 1 according to the present embodiment will be described. In the following, the existing object information generation operation and the patrol operation will be described. The patrol operation is an operation in which the mobile robot 1 travels along the movement route and checks for an abnormality. Existing equipment information is generated prior to patrol and used during patrol. Existing equipment information may be generated regularly or irregularly. Existing object information may be generated when the mobile robot 1 is installed. The existing object information may be generated at an appropriate timing, for example, before the start of daily patrol.

  FIG. 9 is a flowchart showing the existing object information generation process. The mobile robot 1 is set to the existing object information registration mode by receiving an input from an operation unit (not shown) or a command from the monitoring center. When this mode is set, the mobile robot 1 receives the signal from the guide detection unit 7 and controls the movement unit 3 by the movement control unit 5 to control the white line tape 101 (FIG. 2A) as the guide unit. Start moving along. The white line tape 101 is detected from the image, and the movement control unit 5 independently controls the rotations of the motors 35 and 37 so that the white line tape 101 always faces a predetermined angle at a predetermined position of the image. For example, the movement control unit 5 operates so that the white line tape 101 is directed vertically in the center of the image. Further, based on the instruction marker 103, the current moving section and the traveling speed are obtained by referring to the route information 71 in the storage unit 15, and the motor rotation is controlled according to the traveling speed.

  When the mobile robot 1 starts to move, the obstacle detection unit 11 is operated to spatially scan the front. When detecting the reflected light, the obstacle detection unit 11 outputs the distance measurement data S (Φm) of the measurement point. Here, Φm is an angle indicating the irradiation direction of the laser beam. For example, the obstacle detection unit 11 scans a range of −90 ° <Φm <90 °, and the front of the mobile robot is 0 °. The obstacle detection unit 11 outputs measurement data at a predetermined short time interval.

  When receiving the input from the obstacle detection unit 11 (S1, Yes), the obstacle determination unit 13 acquires the position (Xr, Yr) and posture θr of the mobile robot 1 on the environmental map from the self-position detection unit 9. (S3). The self-position detecting unit 9 calculates the position and posture of the mobile robot 1 from the detection values of the resolvers 51 and 53, the wheel radius, the distance between the wheels, and the like by a method generally called dead reckoning.

Next, the output result of the measurement point by the obstacle detection unit 11 is projected on the coordinates on the environment map based on the following formula. That is, the position of the measurement point obtained as a relative position with respect to the mobile robot 1 is converted into coordinates (Xs, Ys) on the environment map (S5).
Xs = Xr + S (Φm) cos (Φm + θr)
Ys = Yr + S (Φm) sin (Φm + θr)

  Next, the grid corresponding to the position where the measurement point exists is determined from the coordinates of the measurement point on the environment map, and the voting value is cumulatively added to the corresponding grid in the voting table (S7). At this time, the voting value calculation unit 97 calculates a voting value that is weighted based on the positional relationship between the measurement points of the mobile robot 1 and the detected object to be measured, and the calculated voting value is added by the voting unit 91. Is done. As described above, the weight is inversely proportional to the distance from the mobile robot 1 to the measurement point. Furthermore, a weight inversely proportional to the distance between adjacent measurement points is given. Although two weighting processes are performed here, only one of the weighting processes may be performed.

  Subsequent to step S7, the obstacle determination unit 13 determines whether or not the measurement of all the movement paths is completed (S9). That is, the guide detection unit 7 detects the path end point indicating marker 103 and determines whether the path end point is recognized. If the end point of the route has not been reached (No at S9), the process returns to step S1 and the above processing is repeated.

  When the end point of the route is reached (S9, Yes), the object determination unit 93 performs an object presence / absence determination process for each grid with reference to the voting table (S11). In this determination process, a grid having a voting value equal to or higher than a predetermined threshold is set as “existing”, and a grid below the threshold is set as “existing”. Then, information on the presence / absence of existing objects for each grid is developed on the environment map as existing object information and stored in the storage unit 15 (S13). Here, the presence / absence of an existing object is registered as attribute information of each grid of the environment map, and this attribute information becomes the existing object information. Therefore, the position of the existing object is specified by the existing object information. In the above processing, the obstacle determination unit 13 functions as an existing object registration unit together with the control unit 21. When the existing object information is stored in the storage unit 15, the voting table is cleared, and the existing object information generation process ends.

  The process for generating the existing object information has been described above. In the above processing, the vote value is added to the grid of measurement points using the vote table, and it is determined that there is an existing object on the grid where the vote value has reached or exceeded the threshold value. This process is based on the following findings. A grid from which measurement points are extracted many times (laser light from the laser sensor hits many times) has a high probability that an object exists in the corresponding location, and the grid's cumulative vote value (cumulative vote point) accordingly Also grows. However, measurement points obtained by reflection of laser light due to disturbance or the like are rarely extracted many times in the same grid, and the vote value of the grid is relatively small. And the voting value of the measurement point erroneously obtained by the blur due to movement is also small. Therefore, it is possible to determine that an existing object exists only in a grid having a high probability that an object exists, based on the determination based on the cumulative vote value as described above.

  In addition, in the above-described existing object information generation process, the influence of disturbance can be further reduced by performing the process of removing the isolated existing object grid. The existing property grid is a grid of existing property “Yes” whose voting value is equal to or greater than the threshold value, and an isolated existing property grid is an existing property grid that is not adjacent to other existing property grids. . Usually, an object is measured with a size larger than a predetermined size, and the influence of environmental noises such as fine objects such as grass and branches, snow, rain, etc. measured only on a single grid by this processing. Can be removed. In the present embodiment, an isolated existing grid is obtained by determining whether the number of existing grids within a predetermined range is equal to or less than a predetermined number (one or more) in the object determination process. The corresponding existing grid determined to be isolated is deleted from the existing information.

  Next, FIG. 10 is a flowchart showing processing when performing abnormality detection by patrol. The process in FIG. 10 is executed after the existing object information is generated.

  The mobile robot 1 is set to the patrol mode when it detects the arrival of a preset patrol start time by a timing means (not shown), or receives an input from the operation unit or a command from the monitoring center, and detects the guide. With the signal from the unit 11 as an input, the movement control unit 5 controls the moving unit 3 to start moving along the white line tape 101 as a guide unit. The traveling control may be the same as when generating an existing object.

  The process (S21, S23, S25) until the measurement point is projected on the environment map by the input from the obstacle detection unit 11 is the same as (S1, S3, S5) in the existing object information generation procedure. Is omitted. Next, detection determination of an intruding object newly appearing in the monitoring area is performed (S27). Details of the intruder determination process will be described later. In the intruder determination process, if there is an intruder, it is determined that there is an abnormality. If there is an intruder (S29, Yes), the mobile robot 1 stops traveling (S35), and further outputs an abnormal signal from the communication unit 17 to the monitoring center (S37).

  If there is no intruder in step S29, detection determination such as disappearance (theft or destruction) of what was in the monitored area is performed (S31). Details of the lost object determination process will also be described later. In the lost object determination process, if there is a lost object, it is determined that there is an abnormality. If there is a lost object (S33, Yes), the mobile robot 1 stops traveling (S35) and outputs an abnormal signal to the monitoring center from the communication unit 17 (S37), as in the case where there is an intruder. .

  If there is no lost object in step S37, it is determined whether or not all routes have been visited (S39). If the route is completed, the voting table is cleared, all the processes are terminated, and the standby mode is entered.

  FIG. 11 is a flowchart showing details of the intruder determination process (appearance object determination process) in step S27 of FIG. 11 is performed by the voting unit 91, the object determining unit 93, and the abnormality determining unit 95 of the obstacle determining unit 13.

  In FIG. 11, first, similarly to the generation of existing object information, the corresponding target grid is determined from the coordinates on the environment map of the measurement point, and the vote value is cumulatively added to the corresponding grid in the voting table (S51). At this time, the weighting is performed in the same manner as when the existing object information is generated, that is, a weighted vote value is calculated, and the vote value is added. As described above, a weight inversely proportional to the distance from the mobile robot 1 to the measurement point is given, and further, a weight inversely proportional to the distance between adjacent measurement points is given. Although two weighting processes are performed here, only one of the weighting processes may be performed.

  Next, referring to the existing object information, if there is an existing object in the target grid to which the vote value is added, the vote value of the target grid on the vote table is set to 0 (S53). Thus, existing objects are not detected when an object is detected from the vote value. Therefore, by providing this step, a process for determining that there is an abnormality is realized only when the object determination unit determines that an object exists in a grid different from the existing object grid.

  In step S53, if there is an existing object in a predetermined grid range (for example, 3 grids × 3 grids) on the environment map centering on the target grid to which the vote value is added, the vote value of the target grid is set to 0. You may do it. This is to prevent false alarms caused by posture fluctuations when the accuracy of the self-position detection unit is not sufficient, and to determine that there is an existing object in the vicinity of the measurement point. belongs to.

  Next, with respect to the target grid to which the vote value is added, it is determined whether there is a grid that already has a vote value among the adjacent grids on the environmental map (S55). If so, they are labeled (S57). That is, grids that have voting values and are adjacent to each other are labeled as one block. Then, the sum of the vote values of the grids belonging to the labeled region is calculated (S59). If the sum of the vote values of the label region is equal to or greater than a predetermined threshold value, the object determination unit 93 determines that an object exists. The abnormality determining unit 95 determines that there is an intruder and there is an abnormality (S61). Since the vote value of a grid with an existing object is set to 0 in step S53, as a result, there is an abnormality only when the object determining unit 93 determines that an object exists in a grid different from the existing object grid. Will be judged.

  In step S55, if there is no adjacent grid having the vote value with respect to the target grid to which the vote value is added, the process proceeds to step S63, and the presence / absence of an intruder is determined based on the vote value of the target grid. Determined by the unit 93. It is determined whether or not the voting value of the target grid is equal to or greater than a threshold value. If the vote value is equal to or greater than the threshold value, it is determined that there is an intruder, and the abnormality determination unit 95 determines that there is an abnormality. Here, since the vote value of the grid with the existing object is set to 0 in step S53, only when the object determining unit 93 determines that an object exists in a grid different from the existing object grid, as in step S61. It is determined that there is an abnormality. Note that the process of step 63 may be omitted, and if there is no adjacent grid having a vote value for the target grid in step S55 (S55-No), the intruder determination process may be terminated. As a result, it is possible to remove the influence of disturbances such as snow and rain measured only on the isolated grid.

  FIG. 6 shows the relationship between each grid, label, and vote value. In the present embodiment, circumscribed rectangles surrounding adjacent grids are obtained, and the grids that enter the circumscribed rectangles are labeled.

  When labeling is performed, not only a stationary object but also a moving object can be detected with a high probability as described below. That is, when the object is stopped, measurement points are obtained each time scanning is performed by the laser sensor, and thus the sum total of grids having a high vote value is calculated. On the other hand, when the object moves, the voting value is given to the grid through which the object passes, so although the voting value of each grid is reduced, the sum of the label areas spread over a wide range is calculated. Is a large value. By performing labeling in this way, there is an advantage that both intruding objects that stop and move can be detected by a simple method. And even when an intruder moves, abnormality detection can be accurately performed. The label area corresponds to the combined area of the present invention, and the function of the combined area extracting unit is realized by the labeling process performed by the obstacle determining unit 13.

  In the present embodiment, the following expansion and contraction processing may be performed as preprocessing for labeling. This process is performed after the vote value of the grid with the existing object is set to 0 in step S53 and before step S55.

  FIG. 7 shows the expansion and contraction process. In FIG. 7A, a grid having voting values in the voting table (a solid grid, here called a voting grid) is developed on the environment map. The “expansion / contraction” process is a general technique in image processing. In FIG. 7B, expansion processing is performed, and in the expansion processing of the present embodiment, eight grids surrounding the voting grid (adjacent grids in eight directions are shown by diagonal lines). (Inflated grid) is added. In FIG. 7C, the contraction process is performed. In the contraction process, the expansion grid at the edge is deleted from the region created by the expansion grid. In FIG. 7D, labeling is performed on the voting grid and the expansion grid after the contraction process. As a result, as shown in the figure, even when the voting grids are adjacent to each other with a non-voting grid (a grid to which no voting value is assigned), the voting grids are collected by labeling. Therefore, adjacent voting grids can be collected by labeling at a reasonable distance set in the expansion process. By such processing, it is possible to absorb measurement point blur due to insufficient accuracy of the self-position detecting unit 9. That is, even if the voting grid is not in direct contact due to measurement point errors or the like, if it is adjacent as shown in the figure (if it is within a predetermined range), extraction by labeling can be performed.

  In the process of FIG. 11, a threshold value is used to determine an intruder, and this threshold value is compared with the voting value in the voting table. This threshold value is set larger than the threshold value used in the existing object determination when generating the existing object information. This setting is for the following reason. If the thresholds at the time of existing object generation and abnormality determination are the same, the sensitivity of both object detections is the same. In this case, it may occur frequently that an existing object is determined as an intruder. Therefore, in the present embodiment, the threshold value at the time of abnormality determination is set large, whereby the sensitivity of abnormality detection can be adjusted appropriately and erroneous detection can be reduced.

  The intruder determination process in step 27 of FIG. 10 has been described above. Next, the lost object determination process in step S31 of FIG. 10 will be described.

  FIG. 12 shows a detailed processing flow of lost object determination. The process of FIG. 12 is performed by the abnormality determination unit 95. First, on the environment map, a measurement vector connecting the mobile robot 1 and the measurement points by the obstacle detection unit 11 is generated (S71). In the present embodiment, a vector connecting the obstacle detection unit 11 of the mobile robot 1 and the measurement point is generated, and more specifically, a vector connecting the position coordinate of the laser sensor 61 and the measurement point is generated.

  Next, a grid through which the measurement vector passes is extracted on the environmental map, starting from the end on the obstacle detection unit 11 side (S73). In step S75 and subsequent steps, the extracted grid is sequentially traced from the starting point, and the presence / absence of a grid (existing object grid) that is set to “existing” in the existing object information is searched.

  In step S75, it is determined whether or not the Nth grid is an existing object grid. If it is not an existing object, it is determined whether or not the Nth grid is the search end point (S77). The search end point is set a predetermined distance before the measurement point (end of the measurement vector). The predetermined distance may be set by the number of grids. If step S77 is No, N = N + 1 is set (S79), the process returns to step S75, and the next grid is determined. If Step S77 is Yes, the existing object grid is not searched and the search end point has been reached (see FIG. 8A). In this case, the abnormality determination unit 95 determines that there is no lost object, that is, no abnormality (S87), and ends the lost object determination process.

  If the existing grid is detected in step S75, it is determined whether or not the Nth grid is the search end point (S81). If the Nth grid is the search end point, the process proceeds to step S87, where the abnormality determination unit 95 determines that there is no lost object, that is, no abnormality, and ends the lost object determination process.

  If step S81 is No, N = N + 1 is set (S83), and it is determined whether or not the Nth grid is an existing grid (S85). If step S85 is Yes, the process returns to step S81. If step S85 is No, the abnormality determination unit 95 determines that there is a lost object, that is, there is an abnormality (S89), and ends the lost object determination process.

  According to the above processing, when an existing object grid is detected, the grid is further searched along the measurement vector, and a grid that is not an existing object grid is searched for. When a grid that is not an existing object grid is detected after the existing object grid is detected (FIG. 8B), the measurement vector penetrates the existing object. It means that there is no longer what exists. In this case, step S85 is No and it is determined that there is a lost item.

  On the other hand, when the existing object grid continues to the end point of the search, the process proceeds from step S81 to step S87, and it is determined that there is no lost object. This determination process is performed in order to prevent erroneous detection, as will be described below with reference to FIG.

  In the present embodiment, the mobile robot 1 is also used for generating existing object information. For this reason, a blur at the time of movement is included in the existing object information, and the existing object may have a thickness (FIG. 8C). For example, the surface of a wall that is an existing object is actually a line on a two-dimensional plane. However, the wall surface may have a thickness due to movement blur. That is, on the detection data, a grid of a band-like portion having a certain width is registered as an existing object.

  Therefore, in this case, the lost object determination is performed using the existing object information having the above thickness. Therefore, as shown in FIG. 8C, a point inside the existing object may be detected as a measurement point. In this case, if it is determined that the existing object has disappeared due to the existing object grid on the measurement vector, an erroneous determination is made. Therefore, in the present embodiment, as described above, when the existing object grid continues to the search end point, the process proceeds from step S81 to step S87, and it is determined that there is no lost object. In this way, erroneous determination is prevented in consideration of the error of measuring existing objects.

  In the present embodiment, as described above, the search range is set from the center of the laser sensor to a point (for example, 15 cm) shorter than the measurement point by a predetermined distance. This also prevents erroneous detection as described below.

  In FIG. 8D, although the measurement point penetrates the existing object grid, it exists in the vicinity of the existing object to be originally measured. In this case, although the existing object is actually detected, there is a possibility that the measurement point has penetrated the existing object grid due to the insufficient accuracy of the self-position detecting unit 11 due to the movement blur or the like. That is, there is a possibility that the position of the existing grid and the measurement point in the tour are different due to the influence of errors during the measurement of existing facilities and during the tour. In the present embodiment, since the search end point is set a predetermined distance before the measurement point of the laser sensor, it is determined as normal in such a case, and erroneous detection can be prevented.

  The preferred embodiments of the present invention have been described above. According to the present embodiment, when the relative position of the object to be measured is detected, the vote value is added to the position (grid) on the environment map corresponding to the detected relative position, and the vote value is a threshold value. It is determined that an object is present at the above position. Even if the detection position shifts in the short term due to the movement of the robot, the vote value does not become so large that it exceeds the threshold value in a short period of time, so it is not necessary to determine that an object exists at the shifted position. In this way, by performing the voting process, it is possible to reduce the influence of the displacement of the detection position accompanying the movement of the robot and to prevent erroneous detection.

  Furthermore, in the present embodiment, the voting value is weighted according to a decreasing function that decreases the weight as the distance to the object to be measured increases. Thus, the detected position is shifted as the robot moves. Even when this occurs, by reducing the weight for the distant measurement point where the influence of the angle deviation becomes large, the influence of the deviation of the detection position can be reduced and the erroneous detection can be prevented.

  In the present embodiment, the vote value is calculated by weighting inversely proportional as the decreasing function, and appropriate weighting can be performed to make the vote value smaller at a distant measurement point.

  In this embodiment, the voting value is given a weight according to a decreasing function that reduces the weight as the distance between adjacent measurement points increases. As a result, even if the detection position is displaced due to the movement of the robot, the influence of the displacement of the detection position is reduced by reducing the weight for the measurement points that are adjacent to each other so that the influence of the angle displacement is large. It can reduce and prevent false detection. Further, the vote value is calculated by weighting inversely proportional to the decreasing function, and appropriate weighting can be performed to make the vote value smaller at the measurement points that are more distant from the adjacent points.

  Further, according to the present embodiment, it is determined that there is an abnormality when it is determined that there is an object in a place where there is no existing object, using the existing object information. Therefore, the existing object is determined as an obstacle. Detection can be prevented, and only abnormal objects such as intruders can be detected.

  Further, according to the present embodiment, it is determined that there is an abnormality when there is a position where the existing object should be on the straight line connecting the self position and the measurement point by the processing using the measurement vector, Thereby, it can be determined that an existing object that should exist cannot be detected, and an abnormality occurring in the existing object can be detected.

  Further, according to the present embodiment, the search range is set up to the search end point set before the detected relative position, thereby preventing erroneous detection.

  In addition, according to the present embodiment, it is determined that there is no abnormality when the position where the existing object should be along the straight line to the search end point in the existing object information, thereby preventing erroneous detection. be able to.

  Moreover, according to this Embodiment, the object detection function for abnormality detection can be utilized in order to acquire existing property information, and existing property information can be obtained easily and stored in a memory | storage part. In addition, by setting the threshold value at the time of abnormality detection larger than the threshold value at the time of acquiring existing object information, it is possible to appropriately adjust the sensitivity of abnormality detection and reduce erroneous detection.

  Further, according to the present embodiment, labeling is performed, and regions that are given voting values and that are adjacent to each other on the environment map are extracted as combined regions. And it is determined whether an object exists in a joint area based on the sum total of the vote value given to each field which constitutes a joint area. This makes it possible to detect the presence of a moving object that moves to an adjacent area before the vote value of one area exceeds the threshold.

  Further, according to the present embodiment, it is determined that no object exists at an isolated position even if the vote value is equal to or greater than a threshold value, and erroneous detection of the object can be prevented.

  The preferred embodiments of the present invention have been described above. However, the present invention is not limited to the above-described embodiments, and it goes without saying that those skilled in the art can modify the above-described embodiments within the scope of the present invention.

  The present invention is useful as a mobile robot that detects surrounding objects while moving.

It is a functional block diagram of the mobile robot which concerns on embodiment of this invention. It is a figure which shows the environment where a mobile robot is used, and the external appearance of a mobile robot. It is a figure which shows the environmental map and existing property information which are memorize | stored in the memory | storage part. It is a figure which shows the influence of the measurement distance with respect to the shift | offset | difference of a measurement point. It is a figure which shows the influence of the measurement direction with respect to the shift | offset | difference of a measurement point. It is a figure which shows a labeling process. It is a figure which shows the labeling process accompanied by an expansion and contraction process. It is a figure which shows a lost article determination process. It is a flowchart which shows the production | generation process of the existing object information. It is a flowchart which shows the process at the time of patrol. It is a flowchart which shows an intruder determination process. It is a flowchart which shows a lost article determination process. It is a figure which shows the weighting process of a voting value.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Mobile robot 3 Moving means 5 Movement control part 7 Guide detection part 9 Self-position detection part 11 Obstacle detection part 13 Obstacle determination part 15 Storage part 17 Communication part 19 Imaging unit 21 Control part 31 Right wheel 33 Left wheel 35, 37 Motor 41 White line detection camera 43 Image processing unit 51, 53 Resolver 55 Position calculation unit 61 Laser sensor 71 Route information 73 Environmental map 75 Existing object information 77 Voting table 79 Current position information 91 Voting unit 93 Object determination unit 95 Abnormality determination unit

Claims (10)

A robot that moves within a predetermined environment by means of traveling means,
A storage unit for storing an environment map corresponding to the predetermined environment;
A position recognizing unit for recognizing a position on the environment map corresponding to the own position;
A detection unit that scans the environment and detects a relative position of the object to be measured;
A voting unit for adding a voting value to a position on the environmental map corresponding to a relative position where the object to be measured is detected by the detection unit;
A voting value calculation unit that weights the voting value by the voting unit based on the relative position of the object to be measured detected by the detection unit;
The object determination unit that determines that an object is present at a corresponding position on the environmental map when the accumulated vote value is equal to or greater than a predetermined threshold; and
The voting value calculation unit calculates a voting value by weighting according to a decreasing function that decreases the weight as the distance to the object to be measured detected by the detection unit increases.   The mobile robot according to claim 1, wherein the voting value calculation unit calculates the voting value by weighting inversely proportional to the distance to the measurement object detected by the detection unit.   The voting value calculating unit further calculates the voting value by weighting according to a decreasing function that decreases the weight as the distance between the relative positions of the adjacent measured objects detected by the detecting unit increases. The mobile robot according to claim 1 or 2.   The mobile robot according to claim 3, wherein the vote value calculation unit calculates the vote value by weighting inversely proportional to the distance between the relative positions of adjacent objects to be measured. Furthermore, the storage unit stores existing object information including position information of an existing object in the predetermined environment,
An abnormality determination unit that determines that there is an abnormality when the object determination unit determines that an object exists at a position different from a position on the environment map corresponding to the existing object information. Item 5. The mobile robot according to any one of Items 1 to 4.   The abnormality determination unit includes a position on the environment map corresponding to a relative position of the object to be measured detected by the detection unit, and a position of the self on the environment map recognized by the position recognition unit. 6. The mobile robot according to claim 5, wherein it is determined that there is an abnormality when a position on the environment map corresponding to the existing object information exists on a connecting line.   The mobile robot according to claim 6, wherein a search range of a position corresponding to the existing object information on the straight line is set from the self position to a position a predetermined distance before the relative position. The storage unit stores a voting table that divides the environmental map into a plurality of regions and stores voting values for each region;
The voting unit adds a voting value to the region corresponding to the relative position of the object to be measured detected by the detection unit, and stores the voting value in the voting table,
The object determination unit determines that an object exists in the corresponding area when a vote value of the voting table is equal to or greater than a predetermined threshold;
And a combined region extracting unit that extracts regions adjacent to each other on the environmental map as the combined region in the region to which the vote value is given,
The object determination unit determines whether or not an object exists in a combined region based on a sum of voting values given to each region constituting the combined region extracted by the combined region extraction unit. Item 8. The mobile robot according to any one of Items 1 to 7.   The object determination unit determines that no object exists in the corresponding area when the number of positions where the vote value is equal to or greater than a threshold value is equal to or less than a predetermined number within a predetermined range. The mobile robot according to any one of 8. A storage unit for storing an environment map corresponding to a predetermined environment;
A detection unit that scans a predetermined scanning range and detects a relative position of the object to be measured;
A voting unit for adding a voting value to a position on the environmental map corresponding to the relative position of the object to be measured detected by the detection unit;
A voting value calculation unit that weights the voting value by the voting unit based on the relative position of the object to be measured detected by the detection unit;
The object determination unit that determines that an object is present at a corresponding position on the environmental map when the accumulated vote value is equal to or greater than a predetermined threshold; and
The voting value calculation unit calculates the voting value by weighting according to a decreasing function that decreases the weight as the distance from the object to be measured detected by the detection unit increases.