JP5720292B2 - Estimated position evaluation system and program - Google Patents

Description

  The present invention relates to an estimated position evaluation system and program for evaluating the accuracy of a position on an estimated environment map generated by self-position estimation with respect to a position on an actual environment map.

  SLAM (Simultaneous Localization And Mapping) is a method of simultaneously performing self-position estimation and creation of an environmental map. However, it is difficult to associate the estimated position on the estimated environment map generated by SLAM with the actual position on the actual environment map because these two environment maps are different.

  FIG. 1 is a diagram for explaining an actual environment map and an estimated environment map generated by SLAM. In FIG. 1, (a) shows a real environment map 1 including an area 2 where a robot or the like can move, and (b) shows an estimated environment map 11 including an area 12 where the robot or the like can move. Regions 2 and 12 to which the robot or the like can move are indicated by hatching in FIG. As shown in FIG. 1, in the estimated environment map 11 generated by SLAM, distortion occurs between the estimated environment map 11 and the actual environment map 1 due to a deviation in the posture of the robot at the initial position of the robot, a measurement error at the time of measurement, and the like. For example, if the robot position at the start of SLAM is the origin and is indicated by a mark ● in FIG. 1, this robot position is the origin on both the estimated environment map 11 generated by SLAM and the actual environment map 1. However, at a position moved, for example, 1 m from the origin to the front of the robot, there is a shift between the robot position on the estimated environment map 11 generated by SLAM and the robot position on the real environment map 1. This deviation increases as the robot position moves away from the origin. For this reason, it is difficult to evaluate how much the self-position estimated by the robot at an arbitrary point on the estimated environment map 11 is different from the actual position on the actual environment map 1.

  Therefore, for example, the front direction at the initial position of the robot is defined as the x-axis (m) and the left-hand direction is defined as the y-axis (m), for example, coordinates (1,0), (2,0), (2,1), Compare the robot position on the actual environment map in (2, 2) with the robot position on the estimated environment map generated by SLAM. A method for mapping to a position on a map has also been proposed. FIG. 2 is a diagram for explaining a method for mapping from a position on the actual environment map to a position on the estimated environment map. In FIG. 2, (a) shows the actual environment map 1, (b) shows the estimated environment map 11, the same parts as those in FIG.

  However, when performing such mapping, if the number of positions to be mapped is small, how much the self-position estimated by the robot at any point on the estimated environment map 11 deviates from the actual position on the actual environment map 1. The accuracy of evaluation is reduced. On the other hand, if the number of mapping positions is large, a decrease in evaluation accuracy can be suppressed, but the mapping process becomes complicated and the processing time increases.

JP 2004-139265 A JP 2004-294421 A

  Conventionally, there has been a problem that it is relatively easy and accurate to evaluate how much the self-position estimated at an arbitrary point on the estimated environment map deviates from the actual position on the actual environment map. It was.

  Therefore, the present invention can relatively easily and accurately evaluate how much the self-position estimated at an arbitrary point on the estimated environment map is deviated from the actual position on the actual environment map. An object is to provide an estimated position evaluation system and program.

According to an aspect of the present invention, there is provided an estimated position evaluation system for evaluating self-position estimation of a robot, wherein the robot position on a real environment map when the robot passes between feature point pairs sandwiching a movement path is determined. Storage means for receiving a self-position estimation result representing the result from the robot and storing it in a storage unit; coordinate values on the estimated environment map of the feature point pair stored in the storage unit; and Calculation means for calculating a robot position on the estimated environment map based on coordinate values on the estimated environment map of the robot when passing between the robot, the robot position on the estimated environment map, and Evaluation means for comparing the self-position estimation results stored in the storage unit and evaluating that the self-position estimation accuracy is higher as the self-position estimation results are closer to the robot position on the estimated environment map. Estimated position evaluation system with is provided.

According to one aspect of the present invention, there is provided a program for causing a computer to evaluate a self-position estimation of a robot, which represents a robot position on a real environment map when the robot passes between feature point pairs sandwiching a movement path. A storage procedure for receiving a self-position estimation result from the robot and storing it in the storage unit; a coordinate value on the estimated environment map of the feature point pair stored in the storage unit; and A calculation procedure for calculating the position of the robot on the estimated environment map based on the coordinate values on the estimated environment map of the robot when passing between, the position of the robot on the estimated environment map, and the storage The self-position estimation results stored in the unit are compared, and the self-position estimation results are evaluated to be more accurate as the self-position estimation results are closer to the robot position on the estimated environment map. Program instructions Ru said to the computer running is provided.

  According to the disclosed estimated position evaluation system and program, it is relatively easy and accurate how much the self-position estimated at an arbitrary point on the estimated environment map deviates from the actual position on the actual environment map. It becomes possible to evaluate well.

It is a figure explaining the presumed environment map produced | generated by a real environment map and SLAM. It is a figure explaining the method of mapping from the position on a real environment map to the position on an estimated environment map. It is a block diagram which shows an example of a structure of the server apparatus in one Example of this invention. It is a block diagram which shows an example of a structure of the robot in one Example of this invention. It is a figure which shows an example of the presumed environment map produced | generated by SLAM. It is a flowchart explaining the 1st example of a position accuracy evaluation process. It is a flowchart explaining the process which determines a feature point pair. It is a figure explaining the process of step S10. It is a flowchart explaining the process of step S10 in detail. It is a flowchart explaining the 2nd example of a position accuracy evaluation process. It is a flowchart explaining the 3rd example of a position accuracy evaluation process. It is a figure explaining the process of step S25. It is a flowchart explaining the process of step S25 in detail. It is a flowchart explaining the 4th example of a position accuracy evaluation process.

  The disclosed estimated position evaluation system and program evaluate self-position estimation of a robot. Obtain the self-position estimation result when the robot passes between the feature point pairs across the movement path, and the coordinate value on the estimated environment map of the feature point pair and the robot's time when the robot passes between the feature point pairs The robot position on the estimated environment map is calculated based on the position information based on the coordinate values. The accuracy of self-position estimation is evaluated by comparing the robot position on the estimated environment map with the self-position estimation result.

  Embodiments of the disclosed estimated position evaluation system and program will be described below with reference to the drawings.

  The estimated position evaluation system in an embodiment of the present invention can be formed by one server device and at least one robot that can communicate with each other via a network.

  FIG. 3 is a block diagram showing an example of the configuration of the server device in one embodiment of the present invention. The server device 20 shown in FIG. 3 has a configuration in which a CPU (Central Processing Unit) 21, a storage unit 22, an input unit 23, a display unit 24, and a communication unit 25 are connected by a bus 29. Note that the CPU 21 may be directly connected to at least one of the storage unit 22, the input unit 23, the display unit 24, and the communication unit 25 without using the bus 29.

  The CPU 21 forms a processor that controls the entire server device 20. The storage unit 22 is formed of a semiconductor storage device, an optical storage device, a magneto-optical storage device, a magnetic storage device, or the like, and may be formed of a plurality of storage devices or a plurality of types of storage devices. The semiconductor storage device includes a RAM (Random Access Memory), a ROM (Read Only Memory), a memory card, and the like. Optical, magneto-optical, and magnetic storage devices include disk devices and the like. The storage unit 22 may include a portable storage device. The storage unit 22 stores programs executed by the CPU 21, intermediate results of calculations executed by the CPU 21, various data, and the like. The input unit 23 is formed by an input device such as a keyboard and a mouse, and various instructions are used to input data to the CPU 21. The display unit 24 is used to display an operation screen, a processing result of the CPU 21, and the like. The communication unit 25 can communicate with a robot 30 described later with reference to FIG. 4 via a wireless network (not shown), and further can communicate with an external device such as a distance sensor via at least one of a wired network and a wireless network. It may be.

  The CPU 21 implements various functions of the server device 20 by executing a program stored in the storage unit 22. That is, the program stored in the storage unit 22 includes procedures for causing the CPU 21 to execute various processes of the server device 20 including the position accuracy evaluation process. The program may include procedures for executing various processes including a feature point extraction process, a feature point pair determination process, an intermediate point determination process, and the like, which will be described later. The computer-readable storage medium storing the program can be formed by at least one storage device forming the storage unit 22 in this example.

  FIG. 4 is a block diagram showing an example of the configuration of the robot in one embodiment of the present invention. The robot 30 is capable of autonomous traveling (that is, autonomous traveling type) and has a known self-position estimation function. The robot 30 shown in FIG. 4 has a configuration in which a CPU 31, a storage unit 32, an input unit 33, a display unit 34, a communication unit 35, a drive unit 36, and a measurement unit 37 are connected by a bus 39. The CPU 31 may be directly connected to at least one of the storage unit 32, the input unit 33, the display unit 34, the communication unit 35, the drive unit 36, and the measurement unit 37 without using the bus 39.

  The CPU 31 forms a processor that controls the entire robot 30. The storage unit 32 is formed of a semiconductor storage device, an optical storage device, a magneto-optical storage device, a magnetic storage device, or the like, and may be formed of a plurality of storage devices or a plurality of types of storage devices. The semiconductor storage device includes a RAM (Random Access Memory), a ROM (Read Only Memory), a memory card, and the like. Optical, magneto-optical, and magnetic storage devices include disk devices and the like. The storage unit 32 may include a portable storage device. The storage unit 32 stores programs executed by the CPU 31, intermediate results of operations executed by the CPU 31, various data, and the like. The input unit 33 is formed by an input device such as a keyboard, and various instructions are used to input data to the CPU 31. The display unit 34 is used to display an operation screen, a processing result of the CPU 31, a message for the user, and the like. The communication unit 55 can communicate with the server device 20 illustrated in FIG. 3 via a wireless network, and further can communicate with an external device such as a distance sensor via at least one of a wired network and a wireless network. . The drive unit 36 has a known configuration including a motor or the like that enables the robot 30 to autonomously travel. The measurement unit 37 includes an LRF (Laser Range Finder) that measures the distance from the robot 30 to an object such as a wall, a camera that captures an image, an optical detection device that detects a marker, a GPS (Global Positioning System), and the like. May be included. The measurement part 37 will not be specifically limited if it is the structure which can implement | achieve the self-position estimation function which the robot 30 employ | adopts.

  The CPU 31 implements various functions of the robot 30 by executing programs stored in the storage unit 32. That is, in the program stored in the storage unit 32, each procedure for causing the CPU 31 to execute various processes of the robot 30 including an autonomous running process and a self-position estimation process for realizing a self-position estimation function for estimating the position of the robot 30. Is included. The program may include procedures for executing various processes including a feature point extraction process, a feature point pair determination process, an intermediate point determination process, and the like. The computer-readable storage medium storing the program can be formed by at least one storage device forming the storage unit 32 in this example.

  When inputting to the robot 30 by voice, a voice input unit (for example, a microphone) is provided instead of the input unit 33 or in addition to the input unit 33, and the CPU 31 has a voice recognition function. good. Further, in the case where the output from the robot 30 to the user is performed by voice, a voice output unit (for example, a speaker) is provided instead of the display unit 34 or in addition to the display unit 34, and the CPU 31 has a voice synthesis function. You can do it.

  FIG. 5 is a diagram illustrating an example of an estimated environment map generated by SLAM. In FIG. 5, the same parts as those in FIG. 2B are denoted by the same reference numerals, and the description thereof is omitted. In this example, as shown in FIG. 5, two characteristic points (hereinafter referred to as feature points) on the estimated environment map generated by SLAM, in this example, a map with a grid (hereinafter simply referred to as grid map). Focus on A and B as a pair. A plurality of pairs of feature points A and B may be selected, but FIG. 5 shows only one pair for convenience of explanation. The feature points A and B are, for example, wall corners (projections or recesses), desk corners, partition edges, and the like (hereinafter simply referred to as corners) that form the boundary of the region 12 where the robot 30 can move. It corresponds to. In the example of FIG. 5, the feature point A corresponds to the concave portion of the corner of the wall in the real environment, and the feature point B corresponds to the convex portion of the corner of the wall in the real environment. The feature points A and B can be determined from the LRF measurement result of the measurement unit 37, for example. The coordinates of the feature points A and B on the grid map are obtained, and the position of the point between the feature points A and B (for example, the midpoint of the feature points A and B) is obtained. When evaluating the accuracy of the self-position estimation process of the robot 30, the robot 30 is allowed to pass through the points between the feature points A and B. When the robot 30 is located at a point between these feature points A and B, the difference between the robot position estimated from the self-position and the point position between the feature points A and B calculated from the grid map is examined. Evaluate the accuracy of the estimation. It can be evaluated that the smaller the difference between the robot position estimated by the self-position and the position of the point between the feature points A and B calculated from the grid map, the higher the accuracy of the self-position estimation.

  Therefore, the position on the estimated environment map generated by SLAM can be associated with the position on the real environment map. For this reason, it is possible to evaluate how much the position estimated by the robot 30 at a certain point is different from the position on the estimated environment map generated by SLAM.

  In the following description, for example, a corner where a characteristic point such as a corner of a wall (convex portion or concave portion), a corner of a desk, or an edge of a partition is provided is referred to as a “corner”.

  FIG. 6 is a flowchart for explaining a first example of the position accuracy evaluation process. The position accuracy evaluation process shown in FIG. 6 is executed by the CPU 21 of the server device 20, for example.

  In FIG. 6, in step S <b> 1, based on the data received from the robot 30 by the communication unit 25, grid map data generated by SLAM is acquired by a known method and stored in the storage unit 22. In step S2, it is determined whether or not the feature point pair is determined manually. Whether the feature point pair is determined manually may be selected by the input unit 23 or may be determined in advance by default. If the decision result in the step S2 is YES, the process advances to a step S5 described later. If the decision result in the step S2 is NO, a step S3 detects a corner existing at the boundary of the movable region 12 of the robot 30 from the grid map data by a known method and stores it in the storage unit 22. For example, the grid map data may be stored in advance in the storage unit 22, input from the input unit 23, or received from the robot 30 by the communication unit 25. In step S4, adjacent feature point pairs that sandwich the movement path of the robot 30 are determined from the plurality of corners detected in step S3 and stored in the storage unit 22.

  FIG. 7 is a flowchart illustrating the process of step S4 for determining a feature point pair. In FIG. 7, step S401 obtains a corner detected from grid map data. In step S402, the movement path of the robot 30 is set for the grid map data. The movement path of the robot 30 is input from the input unit 23, for example. In step S 403, for example, the left side (or right side) corner A of the set travel route is detected along the travel route and stored in the storage unit 22. In step S <b> 404, a corner that is a candidate for a pair is detected for the detected corner A and stored in the storage unit 22. In step S <b> 405, the corner B that minimizes the distance to the corner A is selected from the corners that are pair candidates, and is stored in the storage unit 22. In step S406, if the distance between the corner A and the corner B is equal to or smaller than the reference value, the corners A and B are determined as a feature point pair and stored in the storage unit 22. In step S407, it is determined whether or not the processing in steps S403 to S406 has been executed for all corners detected along the left side of the movement route. If the determination result is NO, the processing returns to step S403. On the other hand, if the decision result in the step S407 is YES, the process advances to a step S5 shown in FIG.

  When the determination result of step S2 is YES and the feature point pair is determined manually, the operator (or the administrator) visually sees two corners on the grid map displayed on the display unit 24, for example. A feature point pair that is a feature point and that the robot 30 passes between these two feature points is selected from the input unit 23 and determined. In FIG. 6, step S5 is automatically determined for the feature point pair determined manually if the determination result in step S2 is YES, or in steps S3 and S4 if the determination result in step S2 is NO. The coordinate values on the grid map are obtained for the feature point pairs, and the process proceeds to step S6.

  In step S6, the distance sensor 900 is attached to a position on the real environment map (that is, a corner of the real environment) corresponding to the feature point pair on the grid map. For example, when feature points A and B indicated by circles in FIG. 5 are selected, the distance from the feature point A to the feature point B, the distance from the feature point B to the feature point A, As shown in FIG. 5, the distance measuring sensor 900 is attached so as to measure both of these distances. For the distance sensor 900, for example, an ultrasonic sensor, an infrared sensor, an LRF, or the like may be used. Therefore, the distance sensor 900 detects the distance to the robot 30 that moves along the moving path in the real environment, and distance data indicating the detected distance (that is, based on the coordinate values of the feature points A and / or B). Distance information) is transmitted to the server device 20 via the network. The server device 20 stores the distance data from the distance sensor 900 received by the communication unit 25 via the network in the storage unit 22. When the distance sensor 900 is provided for each of the feature points A and B, the CPU 21 of the server device 20 uses the outputs of the two distance sensors 900 so that only one distance sensor 900 is provided. The position of the robot 30 on the moving path can be detected more accurately than in the case where there is not. The output of the distance sensor 900 may be transmitted to the robot 30 via a network. In this case, the robot 30 stores the distance data from the distance sensor 900 received by the communication unit 35 via the network in the storage unit 32 and transmits the data from the communication unit 35 to the server device 20 via the network as necessary. May be.

  After step S6, the operator places the robot 30 at a position corresponding to a predetermined position on the grid map generated by the SLAM, and starts the self-position estimation program for the robot 30. After the self-position estimation program is started and the self-position estimation process of the robot 30 is started, the initial position of the robot 30 is set at the corresponding coordinate position on the grid map generated by SLAM. The activation of the self-position estimation program of the robot 30 and the setting of the initial position may be performed by an operator inputting an instruction to the robot 30 from the input unit 23 of the server 20 and transmitting the instruction from the communication unit 25 to the robot 30. The operator may directly input an instruction to the robot 30 from the input unit 33 of the robot 30. Thereafter, in step S7, an autonomous traveling instruction for causing the robot 30 to perform autonomous traveling (that is, autonomous traveling processing) is transmitted via the communication unit 25, and the driving unit 36 of the robot 30 is a well-known method based on the autonomous traveling instruction. The robot 30 moves along the movement path. In step S <b> 8, the robot 30 receives the self-position estimation result of the robot 30 when the robot 30 passes between the feature point pairs determined as described above, and stores the result in the storage unit 22. The self-position estimation result of the robot 30 is obtained within the robot 30 and transmitted to the server device 20. In step S9, the distance data transmitted from the distance sensor 900 when the robot 30 passes between the feature point pairs determined as described above is stored in the storage unit 22. The distance data may be received directly by the server device 20 from the distance sensor 900 or via the robot 30. In step S10, the robot on the grid map calculated based on the position of the feature point pair on the grid map stored in the storage unit 22 and the distance data from the distance sensor 900 stored in the storage unit 22 The accuracy of self-position estimation is evaluated by comparing the position of 30 with the self-position estimation result of the robot 30 stored in the storage unit 22, and the process ends. It can be evaluated that the accuracy of the self-position estimation result is higher as it is closer to the position of the robot 30 on the grid map.

  FIG. 8 is a diagram for explaining the processing in step S10 shown in FIG. In FIG. 8, the same parts as those in FIG. In the example shown in FIG. 8, the coordinate value (x1, y1) of the feature point A on the grid map of the feature point pair is, for example, (2, 2), and the coordinate value (x2) of the feature point B on the grid map. , y2) is, for example, (3, 3). In FIG. 8, C indicates the passing position of the robot 30. Although x and y in the coordinate values (x, y) on the grid map are arbitrary units in this example, it goes without saying that units such as meters (m) may be used.

  FIG. 9 is a flowchart for explaining the process of step S10 in more detail. In FIG. 9, when the robot 30 passes between the feature point pairs A and B, the distance data (from the feature point A to the passing position C of the robot 30) is detected by the distance sensor 900 attached to the feature point A (step S1001). For example, 0.8) is received from the distance sensor 900 and stored in the storage unit 22. In step S1002, when the robot 30 passes between the feature point pairs A and B, the self-estimated position (for example, (2.7, 2.8)) obtained by the self-position estimation process of the robot 30 is received from the robot 30 and stored. Stored in the unit 22. In step S1003, for example, the coordinates of the passing position C of the robot 30 are calculated from the distance data (for example, 0.8) received directly from the distance sensor 900 and stored in the storage unit 22, and stored in the storage unit 22. The coordinates of the passing position C of the robot 30 are the coordinate value of the feature point A (for example, (2, 2)), the unit vector in the direction of the vector AB (for example, (0.707, 0.707)), and the distance data received from the distance sensor 900 ( For example, it can be calculated by adding the product of 0.8). In the example shown in FIG. 8, the coordinate value of the passing position C obtained as a result of the calculation is (2.6, 2.6). In step S1004, the difference between the coordinates of the self-estimated position of the robot 30 (for example, (2.7, 2.8)) stored in the storage unit 22 and the calculated coordinates of the passage position C of the robot (for example, (2.6, 2.6)) is calculated. The obtained value is used as an evaluation value for the accuracy of self-position estimation, and the process ends.

  Although only one feature point pair is shown in FIG. 5 used in the above example, it goes without saying that the accuracy of self-position estimation may be evaluated by determining a plurality of feature point pairs.

  FIG. 10 is a flowchart illustrating a second example of the position accuracy evaluation process. The position accuracy evaluation process shown in FIG. 10 is executed by the CPU 21 of the server device 20, for example. 10, the same steps as those in FIG. 6 are denoted by the same reference numerals, and the description thereof is omitted.

  In the first example of the position accuracy evaluation process, the distance sensor 900 is attached to the position of the feature point pair to evaluate the accuracy of self-position estimation. In the second example, the distance is included in the measurement unit 37 of the robot 30. The distance between the robot 30 and the feature point is measured using the LRF to evaluate the accuracy of self-position estimation. For this reason, it is not necessary to attach the distance sensor 900 to the position of the feature point pair.

  In FIG. 10, after step S5, step S16 transmits an autonomous traveling instruction for causing the robot 30 to autonomously travel via the communication unit 25, and the driving unit 36 of the robot 30 is a well-known method based on the autonomous traveling instruction. The robot 30 moves along the movement path. In step S17, a corner detection instruction for performing corner detection in the actual environment while the robot is autonomously traveling is transmitted via the communication unit 25, and corner detection in the actual environment is performed by the LRF mounted on the robot 30. When the corner is detected, the server device 20 receives the detected corner information and the self-position estimation result of the robot 30 from the robot 30, and the position of the robot 30 estimated by the self-position (that is, the self-position estimation result). ) To calculate the position of the corner on the grid map. In step S18, it is determined whether or not the detected corner matches the feature point pair obtained in step S5. Specifically, if the calculated position of the corner on the grid map is sufficiently close to the coordinate value of the feature point pair obtained in step S5, it is determined that the detected corner matches the feature point pair. If the decision result in the step S18 is NO, the process returns to the step S16.

  On the other hand, if the decision result in the step S18 is YES, a step S19 receives the self-position estimation result when the robot 30 passes between the feature point pairs determined as described above from the robot 30, and stores the storage unit 22 To store. In step S20, the distance data of the distance L between the robot 30 and the feature point pair when the robot 30 passes between the feature point pairs determined as described above (that is, the coordinate values of the feature points A and / or B are obtained). Reference distance information) is received from the LRF mounted on the robot 30 and measuring the distance L, and stored in the storage unit 22. In step S21, the position of the robot 30 on the grid map calculated based on the position of the feature point pair on the grid map stored in the storage unit and the distance data of the distance L stored in the storage unit 22 Is compared with the self-position estimation result of the robot 30 received from the robot 30 and stored in the storage unit 22, the self-position estimation accuracy is evaluated, and the process ends.

  In this case as well, it goes without saying that the accuracy of self-position estimation may be evaluated by determining a plurality of feature point pairs.

  FIG. 11 is a flowchart for explaining a third example of the position accuracy evaluation process. The position accuracy evaluation process shown in FIG. 11 is executed by the CPU 21 of the server device 20, for example. In FIG. 11, the same steps as those in FIG. 6 are denoted by the same reference numerals, and the description thereof is omitted.

  In the first example of the position accuracy evaluation process, the robot 30 is caused to autonomously travel. In the third example, the robot 30 is manually moved.

  In FIG. 11, after step S <b> 4, step S <b> 25 determines which points between the determined feature point pairs are allowed to pass through the robot 30. For example, when passing the midpoint of the feature point pair to the robot 30, the midpoint of the feature point pair is calculated as described with reference to FIGS. 12 and 13 and stored in the storage unit 22.

  FIG. 12 is a diagram illustrating the process in step S25. In FIG. 12, the same parts as those in FIG. In the example shown in FIG. 12, the coordinate value (x1, y1) of the feature point A on the grid map of the feature point pair is, for example, (2, 2), and the coordinate value (x2) of the feature point B on the grid map. , y2) is, for example, (3, 3). In FIG. 12, D indicates the passing position of the robot 30, that is, the midpoint of the feature point pair A, B. In this case, the coordinate value (x0, y0) of the midpoint D on the grid map is (2.5, 2.5).

  FIG. 13 is a flowchart for explaining the process of step S25 in more detail. In FIG. 13, step S251 obtains the coordinate value (x1, y1) of the feature point A on the grid map, for example, (2, 2), and step S252 obtains the coordinate value (x2) of the feature point B on the grid map. , y2), for example, (3, 3) is acquired. In step S253, the coordinate value (x0, y0) of the passing position D of the robot 30 is determined from the coordinate values (x1, y1), (x2, y2) of the feature point pair A, B and stored in the storage unit 22.

  Next, preparation work in a real environment is described. In step S26, the robot passing position on a straight line connecting the corner positions on the real environment map corresponding to the feature point pairs A and B on the grid map is determined. For example, when the midpoint D of the feature point pair A, B is determined as the passing position of the robot 30, the step S27 determines the midpoint of the corner position on the real environment map corresponding to the feature point pair A, B, for example, a measuring device. The marker 950 is attached to the midpoint D of the floor surface in the real environment. Here, as the marker 950, a reflective plate such as a mirror is pasted on the floor surface, even if a sticker is pasted on the floor surface, or an RFID (Radio Frequency IDentification) tag is pasted or embedded on the floor surface. It may be attached or embedded. Further, the marker 950 may be pasted or embedded in the ceiling of the real environment. Thereby, the preparation work in the real environment is completed.

  Next, after step S27, the operator places the robot 30 at a position corresponding to a predetermined position on the grid map generated by the SLAM, and activates the self-position estimation program of the robot 30. The predetermined position on the grid map is in the vicinity of the midpoint D between the feature point pairs A and B in this example. After the self-position estimation program is started and the self-position estimation process of the robot 30 is started, the initial position of the robot 30 is set at the corresponding coordinate position on the grid map generated by SLAM. The activation of the self-position estimation program of the robot 30 and the setting of the initial position can be performed in the same manner as in the first example of the position accuracy evaluation process. Thereafter, in step S28, an autonomous traveling instruction for causing the robot 30 to perform autonomous traveling is transmitted via the communication unit 25, and the driving unit 36 of the robot 30 is driven by a known method based on the autonomous traveling instruction. The robot 30 moves along the movement path and passes over the marker 950. In step S <b> 29, the self-position estimation result of the robot 30 when the robot 30 is positioned on the marker 950 is received from the robot 30 and stored in the storage unit 22. In step S30, the position of the midpoint D determined in step S25 and stored in the storage unit 22 (that is, distance information based on the coordinate values of the feature points A and / or B) is acquired and stored in step S29. The accuracy of self-position estimation is evaluated by comparing the self-position estimation results of the robot 30 stored in the unit 22, and the process ends.

  In step S <b> 29, the fact that the robot 30 has passed over the marker 950 can be detected by a well-known method using a detection device provided in the measurement unit 37. For example, the marker 950 is imaged and detected by a camera provided in the measurement unit 37, or a signal from an RFID tag forming the marker 950 is received by a tag reader provided in the measurement unit 37 to detect the RFID tag. The light that is emitted from the light emitting element of the optical detection device provided in the measuring unit 37 and reflected by the reflection plate that forms the marker 950 is received by the light receiving element of the optical detection device, and the reflection plate is detected. Can do. A camera, a tag reader, and an optical detection device are examples of the detection device.

  Note that the robot 30 may be moved manually instead of causing the robot 30 to autonomously travel in step S28.

  Needless to say, a plurality of markers 950 may be provided, and the accuracy of self-position estimation with respect to the position of each marker 950 may be evaluated.

  FIG. 14 is a flowchart illustrating a fourth example of the position accuracy evaluation process. The position accuracy evaluation process shown in FIG. 14 is executed by the CPU 21 of the server device 20, for example. In FIG. 14, the same steps as those in FIG. 6 are denoted by the same reference numerals, and the description thereof is omitted.

  In the first example, the second example, and the third example of the position accuracy evaluation process, the accuracy of the self-position estimation result is evaluated. On the other hand, in the fourth example of the position accuracy evaluation process, the self-position estimation result is corrected by reflecting the position accuracy evaluation result in the self-position estimation result. In FIG. 14, the processing from step S <b> 1 to S <b> 7 is the same as that in the first example of the position accuracy evaluation processing.

  In FIG. 14, after step S7, step S38 acquires the distance data transmitted from the distance sensor 900 when the robot 30 passes between the feature point pairs determined as described above, and stores it in the storage unit 22. To do. The distance data may be received directly by the server device 20 from the distance sensor 900 or via the robot 30. Step S39 is based on the measurement result of the distance sensor 900 when the robot 30 passes between the feature point pairs determined as described above, that is, based on the distance data stored in the storage unit 22, on the grid map. The position of the robot 30 is calculated and stored in the storage unit 22. In step S39, the self-position estimation result of the robot 30 when the robot 30 passes between the feature point pairs determined as described above is received from the robot 30 and stored in the storage unit 22. The self-position estimation result of the robot 30 is obtained within the robot 30 and transmitted to the server device 20. Step S40 evaluates the accuracy of self-position estimation by comparing the position of the robot 30 on the grid map stored in the storage unit 22 with the self-position estimation result of the robot 30 stored in the storage unit 22. For example, if the evaluation result is below a certain accuracy, the self-position estimation result corrected by correcting the self-position estimation result at the position of the robot 30 on the grid map is transmitted to the robot 30, and the evaluation result exceeds the certain accuracy. Then, the process ends without correcting the self-position estimation result. The correction of the self-position estimation result may be performed in the robot 30. Further, the self-position estimation result may be corrected by the position of the robot 30 on the grid map regardless of the evaluation result of the self-position estimation accuracy. In particular, in the configuration in which the self-position estimation result is corrected in the robot 30, if the evaluation result exceeds a certain accuracy, the self-position estimation result is not corrected so that the load on the CPU 31 in the robot 30 is reduced. Can be reduced.

  In FIG. 14, the self-position estimation result obtained by the method of the first example of the position accuracy evaluation process is corrected. Similarly, the self-position obtained by the method of the second example of the position accuracy evaluation process is corrected. The position estimation result is corrected by appropriately changing the processing of steps S19 to S21 shown in FIG. 10, or the self-position estimation result obtained by the method of the third example of the position accuracy evaluation process is shown in FIG. You may correct | amend by changing suitably the process of S29 and S30.

  In the above embodiment, the self-position estimation evaluation and related processing are basically performed on the server device 20 side. However, at least a part of the processing may be performed on the robot 30 side. The processing that can be executed on the robot 30 side includes, for example, feature point extraction processing, feature point pair determination processing, intermediate point determination processing, etc. in addition to self-position estimation processing that realizes a self-position estimation function for estimating the position of the robot 30. Is included. If processing with a relatively large load on the CPU is performed on the server device 20 side, the load on the CPU 31 in the robot 30 can be reduced. For example, as shown in the above embodiment, the self-position estimation process is executed by the CPU 31 in the robot 30 and the other processes are executed by the CPU 21 in the server device 20, thereby increasing the load on the CPU 31 in the robot 30. Can be suppressed.

The following additional notes are further disclosed with respect to the embodiment including the above examples.
(Appendix 1)
An estimated position evaluation system for evaluating self-position estimation of a robot,
Storage means for receiving a self-position estimation result when the robot passes between feature point pairs sandwiching a movement path from the robot and storing it in a storage unit;
The coordinate value on the estimated environment map of the feature point pair stored in the storage unit and the position information based on the coordinate value of the robot when the robot passes between the feature point pair. Calculation means for calculating a robot position on the estimated environment map based on
The robot position on the estimated environment map is compared with the self-position estimation result stored in the storage unit, and the self-position estimation accuracy is closer to the robot position on the estimated environment map. An estimated position evaluation system characterized by comprising evaluation means for evaluating that is high.
(Appendix 2)
The estimated position evaluation system according to claim 1, further comprising means for receiving the position information from a distance sensor that measures a distance from at least one of the feature point pairs to the robot.
(Appendix 3)
The estimated position evaluation system according to appendix 1, further comprising means for receiving the position information from a measurement unit that is provided in the robot and measures a distance to at least one of the feature point pairs.
(Appendix 4)
A determination unit for determining the position information and storing the position information in the storage unit;
The storage means receives the self-position estimation result from the robot when a detection device provided in the robot detects a marker provided between the feature points during the movement of the robot. The estimated position evaluation system according to attachment 1.
(Appendix 5)
The estimated position according to any one of claims 1 to 4, further comprising correction means for correcting the self-position estimation result at the robot position on the estimated environment map based on the evaluation result. Evaluation system.
(Appendix 6)
The estimated position evaluation system according to appendix 5, wherein the correction means corrects the self-position estimation result at the robot position on the estimated environment map when the evaluation result is below a certain accuracy.
(Appendix 7)
The storage means, the calculation means, and the evaluation means are formed by a processor in a server device capable of communicating with the robot,
The estimated position evaluation system according to any one of appendices 1 to 6, wherein the storage unit is provided in the server device.
(Appendix 8)
A program that causes a computer to evaluate the robot's self-position estimation,
A storage procedure for receiving a self-position estimation result when the robot passes between feature point pairs sandwiching a movement path from the robot and storing it in a storage unit;
The coordinate value on the estimated environment map of the feature point pair stored in the storage unit and the position information based on the coordinate value of the robot when the robot passes between the feature point pair. And a calculation procedure for calculating a robot position on the estimated environment map,
The robot position on the estimated environment map is compared with the self-position estimation result stored in the storage unit, and the self-position estimation accuracy is closer to the robot position on the estimated environment map. A program for causing the computer to execute an evaluation procedure for evaluating that the value is high.
(Appendix 9)
The program according to claim 8, further causing the computer to further execute a procedure of receiving the position information from a distance sensor that measures a distance from at least one of the feature point pairs to the robot.
(Appendix 10)
The program according to appendix 8, further comprising causing the computer to further execute a procedure of receiving the position information from a measurement unit that is provided in the robot and measures a distance to at least one of the feature point pairs.
(Appendix 11)
Further causing the computer to execute a determination procedure for determining the location information and storing it in the storage unit;
The storing procedure is characterized in that when the detection device provided in the robot detects a marker provided between the feature points during the movement of the robot, the self-position estimation result is received from the robot. The program according to appendix 8.
(Appendix 12)
The supplementary procedure according to any one of appendices 8 to 11, further comprising: causing the computer to further execute a correction procedure for correcting the self-position estimation result at the robot position on the estimated environment map based on the evaluation result. Program.
(Appendix 13)
13. The estimated position evaluation system according to appendix 12, wherein the correction procedure corrects the self-position estimation result at the robot position on the estimated environment map when the evaluation result is below a certain accuracy.
(Appendix 14)
A computer-readable storage medium storing the program according to any one of appendices 8 to 13.

  The estimated position evaluation system and program disclosed above have been described by way of examples. However, it goes without saying that the present invention is not limited to the above examples, and that various modifications and improvements can be made within the scope of the present invention. Yes.

20 Server device 21, 31 CPU
22, 32 Storage unit 23, 33 Input unit 24, 34 Display unit 25, 35 Communication unit 29, 38 Bus 30 Robot 36 Drive unit 37 Measurement unit

Claims (5)

A program that causes a computer to evaluate the robot's self-position estimation,
A storage procedure for receiving a self-position estimation result representing a robot position on a real environment map when the robot passes between a pair of feature points sandwiching a movement path from the robot and storing it in a storage unit;
The basis of the coordinate values of the putative environmental map of the feature point pairs stored in the storage unit, on the coordinate values on the estimated environmental map of the robot when the robot passes between the feature point pairs A calculation procedure for calculating a robot position on the estimated environment map;
The robot position on the estimated environment map is compared with the self-position estimation result stored in the storage unit, and the self-position estimation accuracy is closer to the robot position on the estimated environment map. A program for causing the computer to execute an evaluation procedure for evaluating that the value is high. When the robot passes between the feature point pairs, the computer further executes a procedure of receiving distance data from a distance sensor that measures a distance from at least one of the feature point pairs to the robot ,
The calculation procedure, and the estimated environmental features that you calculate the coordinate values on the map of the robot coordinate values putative environmental map and on the basis of said distance data of said feature point pairs claim 1, wherein Program. When the robot passes between the feature point pairs, the computer further executes a procedure of receiving distance data from a measurement unit that is provided in the robot and measures a distance to at least one of the feature point pairs ;
The calculation procedure, and the estimated environmental features that you calculate the coordinates on a map of the robot coordinate values putative environmental map and on the basis of said distance data of said feature point pairs, claim 1, wherein Program. 4. The computer according to claim 1, further causing the computer to execute a correction procedure for correcting the self-position estimation result at the robot position on the estimated environment map based on the evaluation result. 5. The listed program. An estimated position evaluation system for evaluating self-position estimation of a robot,
Storage means for receiving a self-position estimation result representing a robot position on a real environment map when the robot passes between feature point pairs sandwiching a movement path from the robot and storing it in a storage unit;
The basis of the coordinate values of the putative environmental map of the feature point pairs stored in the storage unit, on the coordinate values on the estimated environmental map of the robot when the robot passes between the feature point pairs Calculating means for calculating a robot position on the estimated environment map;
The robot position on the estimated environment map is compared with the self-position estimation result stored in the storage unit, and the self-position estimation accuracy is closer to the robot position on the estimated environment map. An estimated position evaluation system characterized by comprising evaluation means for evaluating that is high.

Images