JP2021196489A - Map correction system, and map correction program - Google Patents

Abstract

To provide a map correction system capable of improving accuracy of map data by the reflection of a real space situation.SOLUTION: A map correction system 100 detects differences or errors between maps and real spaces by using a robot self-position estimation result caused by the result of collation between three-dimensional or two-dimensional maps regulating a robot movable range in accordance with a structure in a space and a position of the structure, and, by using a detection result, corrects the maps so that the situation of the structure is reflected.SELECTED DRAWING: Figure 2

Description

The present invention relates to correction of map information necessary for controlling a robot.

Conventionally, a technique related to position estimation of a robot that estimates its own position even at a construction site is known (for example, Patent Document 1). In this technique, mapping information is obtained by extracting from the construction data of a measurable construction object, and the self-position in the space is calculated based on the measurement data and the mapping information.

Further, there is known a technique related to a moving body that can be easily applied even if the layout in the moving space is changed, the introduction cost is low, and the self-position can be recognized accurately at high speed (for example, Patent Document 2). As a similar technique, a technique is known in which a moving body estimates its own position and can autonomously control the movement to a designated destination to stably obtain the position of the moving body (for example, Patent Document 3). ).

Further, in consideration of the circumstances unique to a building construction site, a technique for providing an autonomous mobile device capable of accurately estimating a self-position in the construction site is known (for example, Patent Document 4). With this technology, shape data of multiple target parts existing in the construction site of a building is acquired, drawing data is referred to, and based on information on the dimensions of each of the multiple target parts, from among multiple target parts. The reference target site is specified. In addition, the self-position in the construction site is estimated based on the collation result between the position of the reference target part included in the drawing data and the actual position of the reference target part in the shape data.

JP-A-2018-164966 Japanese Unexamined Patent Publication No. 2010-66934 Japanese Unexamined Patent Publication No. 2019-123544 Japanese Unexamined Patent Publication No. 2020-004231

According to the techniques described in Patent Documents 1 to 4, the self-position of the robot in the real space can be estimated accurately. If the self-position is estimated accurately in this way, it is considered that the robot can move accurately and smoothly on the path. However, there may be a difference between the data on the map showing the route and the data at the actual construction site. This is because various changes occur at the actual construction site, such as changes in the arrangement of materials, etc., and changes in the arrangement due to design changes, depending on the situation. Therefore, it is not possible to guarantee the accuracy of the robot's movement simply by accurately estimating the robot's self-position.

It is an object of the present invention to improve the accuracy of map data by reflecting the situation in real space in consideration of the above facts.

In order to achieve the above object, the map correction system of the present invention collates a three-dimensional or two-dimensional map that defines the movable range of the robot according to the structure in space with the position of the structure by the robot. The difference or error between the map and the real space is detected by using the self-position estimation result by, and the map is corrected so as to reflect the situation of the structure by using the detection result. This makes it possible to improve the accuracy of the map data by reflecting the situation in the real space.

According to the present invention, it is possible to improve the accuracy of the map data by reflecting the situation in the real space.

It is an image diagram of a robot management platform. It is a block diagram which shows the structure of the map correction system which concerns on embodiment of this invention. It is a figure which shows an example of the detection and correction of the difference when an obstacle is removed. It is a figure which shows an example of the detection and correction of the difference when an obstacle is newly installed. It is a figure which shows an example of the detection and correction of an error when an obstacle is moved. It is a sequence diagram which shows the map correction processing in the map correction apparatus which concerns on embodiment of this invention.

[Embodiment of the present invention]
Hereinafter, the map correction system according to the embodiment of the present invention will be described with reference to the drawings.

An outline of the embodiment of the present invention will be described. In the method according to the embodiment of the present invention, the difference or error of the structure between the 3D map and the real space is detected by using the 3D map (or the 2D map) and the self-position estimation result of the robot in the real space. Then, the detection result is used to correct the 3D map. A three-dimensional map is an environmental map in which there is a space in which structures are arranged in a space, and a movable range in which a robot can move in the space is defined. The three-dimensional map is generated by a simulation based on BIM data for each type of robot. The real space is a space where a robot actually operates, such as a construction site.

Further, as a premise of this method, the robot uses a self-position estimation method called SLAM (Simultaneous Localization and Mapping). SLAM is a technology that recognizes the surrounding environment with a robot sensor and at the same time estimates its own position and posture with high accuracy. It estimates its own position and at the same time generates an environmental map. In SLAM, the robot combines a plurality of sensors to recognize the environment. The sensor varies depending on the robot, but a distance measuring sensor, a camera, a gyro sensor, a magnetic sensor, an acceleration sensor, a radar sensor (scanner), a bumper sensor and the like are used. As an example of self-position estimation using a sensor, an example of a four-wheel drive robot (hereinafter referred to as a cleaning robot) for cleaning will be described. For example, the cleaning robot has a laser scanner and a bumper sensor as sensor information. The laser scanner is a sensor that detects surrounding walls and obstacles, and by using the laser scanner, the cleaning area can be determined by detecting the reflection markers placed at the four corners of the area to be cleaned. Further, since the cleaning robot moves in the cleaning area by a route that can clean all the dust in the cleaning area surrounded by the reflective marker, the movement route can be automatically generated in the cleaning area. The bumper sensor is a sensor that detects contact with an obstacle, and detects and avoids an obstacle in the cleaning area.

The robot moves while constantly monitoring the surrounding conditions at its own position by self-position estimation. In self-position estimation, the robot collates the position of a structure in real space with a sensor, and obtains the self-position estimation result from the collation result. Here, a difference or an error may occur between the space of the three-dimensional map and the real space. Specifically, there may be a difference or error between the structure at the construction site in the real space and the structure included in the three-dimensional map obtained by simulation. The structure assumed in space may be a fixed structure or a movable structure. Examples of such movable structures include installations, construction site materials, construction tools and the like. Therefore, at a construction site in a real space, the arrangement of movable structures changes depending on movement, addition, use, and the like. In such a case, a difference or an error occurs between the structure in the real space and the structure contained in the three-dimensional map. The difference is a situation in which the arranged structures are added or removed, and the number and arrangement of the structures in the space are different from those of the three-dimensional map. An error is a situation in which the arrangement of structures arranged on a three-dimensional map is misaligned. In this embodiment, these differences or errors are detected by using a three-dimensional map and a self-position estimation result. Then, the arrangement of the structures on the 3D map is corrected so as to reflect the situation of the structures in the 3D map according to the detection result. This makes it possible to correct the coordinate information of the structure included in the three-dimensional map and improve the accuracy of the map data for controlling the robot. Further, it can be realized not only as the arrangement of the structure as the three-dimensional data captured on the three-dimensional map but also as the accuracy update of the two-dimensional point cloud data. Since SLAM is usually processed in two dimensions, if the map data for robot running extracted from the three-dimensional data of BIM is a two-dimensional map, the two-dimensional data is updated. In addition, the actually detected data is updated according to the actual situation during the construction of the building, while the original BIM data is used after the completion, so the map data to be used is the current data. Or it is managed so that it can be selected from the original data. Along with this, it is possible to improve the accuracy of the robot management system that manages various robots described below.

Here, the robot management platform that is the premise of this embodiment will be described. FIG. 1 is an image diagram of a robot management platform. As shown in FIG. 1, the robot management platform is a platform for managing robots in a cloud environment. Programs that execute functions for managing robots are implemented as various modules on the robot management platform, and the modules are linked as appropriate to perform necessary processing. As a result, the robot management platform realizes automation of construction work by robots. The list of modules shown in FIG. 1 shows only an example of functional means, and is not limited to these examples. By utilizing such a robot management platform, it is possible to eliminate the troublesome setting work for robot operation as much as possible. In addition, by estimating the robot's self-position based on the BIM (Building Information Modeling) data to be constructed, the person in charge at the site can refer to the BIM data and give instructions to the robot, so intuitive operation is possible. Can be provided to field personnel. In addition, functions required for construction, such as monitoring the status of robots by remote control, can be developed as a platform service.

In the robot management platform shown in FIG. 1, the method of this embodiment is used, for example, for SLAM / BIM data linkage. It can also be used for robot operation management, route simulation, data storage / visualization, etc. In the functions of these modules, it is positioned as a method for optimizing the control in real space when the robot moves and improving the performance.

FIG. 2 is a block diagram showing a configuration of a map correction system according to an embodiment of the present invention. As shown in FIG. 2, in the map correction system 100, the map correction device 110, the terminal 140, and the plurality of robots 150 are connected via the network N. The network N is, for example, an internet line, a public wireless LAN, or the like.

The terminal 140 is a terminal that inputs control input data of the robot 150, confirms the self-position estimation result of the robot 150, confirms a corrected three-dimensional map, and the like. The control input data is information including, for example, a destination, a transportation target, and the like. The terminal 140 is a terminal operated by various persons in charge, and performs input / output necessary for processing of the map correction device 110. The various responsibilities here apply to the robot management platform described above. For example, in the present embodiment, the "on-site person in charge" who is in charge of remote control of the robot or the "equipment management person" who is in charge of route simulation is in charge according to the operation. Although the authority is assigned according to the person in charge when the terminal 140 is logged in, the description is omitted here because it is not the main process of the map correction system 100.

The robot 150 is a plurality of robots to be controlled and includes various sensors. Further, the robot 150 receives the control information and moves the route included in the control information while performing self-position estimation. The robot 150 transmits the current position and the self-position estimation result to the map correction device 110 at regular intervals. The robot 150 is mounted as an agent in the simulation environment of the robot environment platform for each type, and can emulate the operation. In this embodiment, a three-dimensional map that defines the movable range is generated in advance by emulating the movement. Further, the route generation of the robot 150 on the three-dimensional map is performed by the route simulation of the route generation unit 114 described later. In this embodiment, a case where a three-dimensional map is used will be described as an example, but as described above, it can be similarly applied to a two-dimensional map.

The map correction device 110 includes a communication unit 112, a route generation unit 114, a detection unit 116, a map correction unit 118, and a storage unit 120. Further, the map correction device 110 can be configured by a computer including a CPU, a RAM, a ROM for storing a program for executing each processing unit, and various data (not shown). The map correction device 110 of the present embodiment is a server constructed by modularizing a part of the functions of the robot management platform of FIG. 1 described above, and each functional unit is an example of the functions.

The communication unit 112 transmits and receives various data by communicating with the terminal 140 and the robot 150. For example, the communication unit 112 receives a three-dimensional map from the terminal 140 and stores it in the storage unit 120. The communication unit 112 receives the control input data of the robot 150 from the terminal 140. The communication unit 112 periodically receives the current position from the robot 150. The communication unit 112 receives the self-position estimation result from the robot 150. Further, the communication unit 112 transmits control information to the robot 150.

Each of the three-dimensional maps received from the terminal 140 is stored in the storage unit 120. The 3D map defines the range of movement of the robot in space. Further, the storage unit 120 stores the route of the robot 150 generated by the route generation unit 114. Further, the corrected three-dimensional map is stored in the storage unit 120.

The route generation unit 114 generates the route of the robot 150 in the three-dimensional map of the storage unit 120 according to the control input data of the robot 150 received from the terminal 140. Then, the control information including the route of the three-dimensional map is transmitted to the robot 150. Further, the control information including the route generated here is information including the route from the starting point to the destination included in the control input data and the necessary work process. The starting point is the latest current position received from the robot 150.

The detection unit 116 detects the difference or error between the three-dimensional map and the real space by using the three-dimensional map of the storage unit 120 and the self-position estimation result received from the robot 150.

The map correction unit 118 corrects the three-dimensional map of the storage unit 120 so as to reflect the state of the structure by using the detection result of the detection unit 116.

Here, specific examples of detection by the detection unit 116 and correction by the map correction unit 118 will be described. FIG. 3 is a diagram showing an example of detection and correction of a difference when an obstacle is removed. In the example shown in FIG. 3, it is assumed that a cleaning area and a non-cleanable area are set as a movable range in the space of the three-dimensional map. The cleaning area corresponds to the path of the cleaning robot and is set in an area where there are no obstacles. In addition, the non-cleanable area is arbitrarily set as an area where obstacles exist and cannot be cleaned. It is assumed that these settings are preset at the stage of generating the three-dimensional map. Here, it is assumed that the cleaning robot is operated in the real space and the difference in the space where the obstacle is removed is detected in the self-position estimation result. In this case, the map correction unit 118 corrects the three-dimensional map by using the detection result of the removed obstacle. In addition, the route generation unit 114 sets a new cleaning area and generates a route. In this way, when an obstacle is newly installed, the three-dimensional map is corrected so that the movable range is widened.

Detection and correction can be performed in the same way when an obstacle is newly installed or when the obstacle is moved. FIG. 4 is a diagram showing an example of detection and correction of a difference when an obstacle is newly installed. As shown in FIG. 4, when an obstacle is newly installed, the three-dimensional map is corrected so that the movable range is narrowed. FIG. 5 is a diagram showing an example of error detection and correction when an obstacle is moved. As shown in FIG. 5, when the obstacle is moved (or is displaced), the 3D map is corrected so as to adjust the movable range. As described above, the real-time situation can be fed back to the three-dimensional map to regenerate the path that the robot can pass.

Next, the operation of the map correction device 110 according to the embodiment of the present invention will be described. FIG. 6 is a sequence diagram showing a map correction process in the map correction device 110 according to the embodiment of the present invention. The map correction process is performed by the CPU reading the program and various data from the ROM and executing the program. The CPU functions as each part of the map correction device 110. It is assumed that the three-dimensional map received from the terminal 140 in advance is stored in the storage unit 120 in advance. In addition, the current location is periodically accepted from the robot 150. Further, it is assumed that the explanation that each functional unit performs the transmission / reception of various information of the map correction device 110 is performed via the communication unit 112.

In step S100, the terminal 140 transmits the control input data of the robot 150 to the map correction device 110. A specific robot 150 is specified in the control input data.

In step S102, the route generation unit 114 generates a route in the three-dimensional map corresponding to the designated robot 150 of the storage unit 120 according to the control input data, and generates control information including the route.

In step S104, the route generation unit 114 transmits control information to the robot 150.

In step S106, the robot 150 moves while performing self-position estimation according to the control information including the received reference area.

In step S108, the robot 150 transmits the self-position estimation result to the map correction device 110.

In step S110, the detection unit 116 detects a difference or an error using the three-dimensional map of the storage unit 120 and the self-position estimation result received from the robot 150.

In step S112, the map correction unit 118 corrects the three-dimensional map of the storage unit 120 by using the detection result of step S110.

In step S114, the route generation unit 114 regenerates the route using the corrected three-dimensional map, and transmits control information including the route to the robot 150. After that, the process proceeds from step S106 to perform map correction processing until the robot 150 arrives at the destination.

As described above, according to the map correction system 100 according to the embodiment of the present invention, it is possible to improve the accuracy of the map data by reflecting the situation in the real space.

The present invention is not limited to the above-described embodiment, and various modifications and applications can be made without departing from the gist of the present invention.

For example, if it is detected that there is a difference or error between the real space and the space of the 3D map, other work is available to make corrections that accurately capture the 3D shape of the structure. You may instruct the robot to make measurements to estimate the shape of the structure. This is because it is assumed that the above-mentioned robot only incidentally captures the structure as a result of self-position estimation in the movement of its own work, and can capture only a part of the situation. Upon receiving the instruction, the robot 150 moves so as to circulate around the difference or the surroundings, and transmits the self-position estimation result to the map correction device 110. The map correction device 110 performs the processing of the detection unit 116 and the map correction unit 118 similar to the above-mentioned control intention, and corrects the three-dimensional map that captures the shape of the structure. As described above, by having other robots perform the work to capture the whole image of the structure in which the difference or error is detected, it is possible to accurately reflect the situation of the structure in the real space on the 3D map. become able to.

100 Map correction system 110 Map correction device 112 Communication unit 114 Route generation unit 116 Detection unit 118 Map correction unit 120 Storage unit 140 Terminal 150 Robot

Claims (4)

Using a three-dimensional or two-dimensional map that defines the movable range of the robot according to the structure in space and the self-position estimation result based on the collation result with the position of the structure by the robot, the map and the real space Detects differences or errors in
Using the detection result, the map is corrected to reflect the situation of the structure.
Map correction system. The map correction system according to claim 1, wherein the correction is made so as to reflect the shape of the structure. The map correction system according to claim 1 or 2, wherein in order to specify the shape of the structure, another robot is instructed to detect the shape of the structure. Using a three-dimensional or two-dimensional map that defines the movable range of the robot according to the structure in space and the self-position estimation result based on the collation result with the position of the structure by the robot, the map and the real space Detects differences or errors in
Using the detection result, the map is corrected to reflect the situation of the structure.
A map correction program that causes a computer to perform processing.

Images