JP2018206038A - Point group data processing device, mobile robot, mobile robot system, and point group data processing method - Google Patents

Abstract

To obtain point group data suitable for accurate map generation even if the amount of the point group data is reduced in order to improve efficiency or increase the speed of integration of the point group data.SOLUTION: A point group data processing device 20 comprises a data amount reduction unit 5. When an environment measurement sensor 3 acquires point group data representing positions of respective coordinate points on an object surface with a set coordinate system, the data amount reduction unit 5 reduces the number of coordinate points included in the point group data by performing data amount reduction processing on the point group data. The data amount reduction processing is a process for setting coordinate point density of short distance point group data to be equal to or less than a first threshold and setting coordinate point density of long distance point group data to be equal to or less than a second threshold, and the first threshold is higher than the second threshold. Alternatively, the data amount reduction processing is a process for reducing the number of coordinate points of the short distance point group data such that the coordinate point density in the short distance point group data is to be equal to or less than the first threshold while keeping the number of coordinate points of the long distance point group data.SELECTED DRAWING: Figure 2

Description

  The present technology relates to a point cloud data processing apparatus and a point cloud data processing method for processing point cloud data acquired by an environment measurement sensor provided in a mobile robot. The present invention also relates to a mobile robot including a point cloud data processing device and a mobile robot system including a plurality of mobile robots.

  The mobile robot includes an environment measurement sensor for recognizing the environment in the vicinity of the mobile robot. The environmental measurement sensor is, for example, a laser scanner. The laser scanner emits laser light in each direction from the mobile robot, receives reflected light from each measurement point on the object surface in the measurement range (for example, around the mobile robot), and obtains the position of the measurement point. . That is, the laser scanner acquires point cloud data representing the position (coordinates) of each measurement point on the object surface.

  SLAM (Simultaneous Localization and Mapping) is known as a technique for autonomously moving a mobile robot based on such point cloud data. In SLAM, based on the point cloud data acquired by the environment measurement sensor of the mobile robot, the posture and position of the mobile robot are estimated by itself, and at the same time, map data indicating the location of surrounding obstacles is generated. This technique is described in Patent Document 1 below, for example.

  In a mobile robot system including a plurality of mobile robots, a plurality of point cloud data are integrated. An environment measurement sensor mounted on the first mobile robot acquires first point cloud data by measurement from the first point. An environmental measurement sensor mounted on the second mobile robot acquires second point cloud data by measurement from the second point. The first mobile robot receives the second point cloud data transmitted from the second mobile robot, and integrates the received second point cloud data into the first point cloud data to generate map data. . Thereby, wide-area map data can be obtained.

  Alternatively, when one mobile robot acquires each point cloud data by measurement from a plurality of different points, the plurality of point cloud data are integrated to generate map data. Thereby, wide-area map data can be obtained.

JP 2009-205226 A

  By the way, it is conceivable to reduce the amount of point cloud data so that the point cloud data can be integrated efficiently or at high speed. For example, it is conceivable to perform a filtering process on the point cloud data to suppress the density of measured points in the point cloud data to a low value. However, high-density measurement points are suitable for generating highly accurate map data. That is, since the density of the measurement points decreases due to the filtering process, there is a possibility that accurate map data cannot be generated.

  Therefore, the purpose of this technology is to obtain point cloud data suitable for generating a highly accurate map even if the amount of point cloud data is reduced in order to increase the efficiency or speed of integration between point cloud data. Is to make it.

In order to achieve the above object, a point cloud data processing device according to an embodiment of the present technology processes point cloud data acquired by an environment measurement sensor. The environmental measurement sensor acquires point group data in which the position of each coordinate point is expressed in a set coordinate system with each measured point on the object surface in the measurement range viewed from the measurement position as a coordinate point.
The point cloud data processing device includes a data amount reduction unit that reduces the number of coordinate points included in the point cloud data by performing a data amount reduction process on the point cloud data.
The point cloud data includes short-distance point cloud data and long-distance point cloud data. In the setting coordinate system, the short-distance point cloud data indicates the position of each coordinate point in the short-distance region on the measurement position side, and The distance point group data indicates the position of each coordinate point in the long-distance area farther from the measurement position than the short-distance area.
Data volume reduction processing
(A) Processing for setting the coordinate point density of the short-distance point group data to be equal to or lower than the first threshold value, and setting the coordinate point density of the long-distance point cloud data to be equal to or lower than the second threshold value, where the first threshold value is higher than the second threshold value. Or (B) the coordinate points of the short-distance point cloud data so that the coordinate point density in the short-distance point cloud data is less than or equal to the first threshold value while maintaining the number of coordinate points of the long-distance point cloud data It is a process to reduce the number of.

  A mobile robot according to an embodiment of the present technology transmits the above-described point cloud data processing device, an environment measurement sensor that acquires point cloud data, and short-distance point cloud data that has undergone data amount reduction processing to another mobile robot. A communication unit is provided.

A mobile robot system according to an embodiment of the present technology includes a plurality of mobile robots.
Each mobile robot
An environmental measurement sensor that obtains point cloud data representing the position of each coordinate point in a set coordinate system, with each measured point on the object surface in the measurement range as seen from the measurement position that is its own position as a coordinate point;
A communication unit that communicates with other mobile robots.
At least one of the mobile robots includes the above-described point cloud data processing device that processes the point cloud data acquired by the environment measurement sensor of the mobile robot.
The communication unit of the mobile robot transmits the short-distance point cloud data after the data amount reduction processing by the point cloud data processing device to another mobile robot, and the other mobile robot performs a map based on the short-distance point cloud data. A map generation unit for generating data is provided.

A point cloud data processing method according to an embodiment of the present technology processes point cloud data acquired by an environmental measurement sensor. When the point cloud data that represents the position of each coordinate point in the set coordinate system is obtained by the environmental measurement sensor with each measured point on the object surface in the measurement range as seen from the measurement position as the coordinate point, The number of coordinate points included in the point cloud data is reduced by performing the data amount reduction process.
The point cloud data includes short-distance point cloud data and long-distance point cloud data. In the setting coordinate system, the short-distance point cloud data indicates the position of each coordinate point in the short-distance region on the measurement position side, and The distance point group data indicates the position of each coordinate point in the long-distance area farther from the measurement position than the short-distance area.
Data volume reduction processing
(A) Processing for setting the coordinate point density of the short-distance point group data to be equal to or lower than the first threshold value, and setting the coordinate point density of the long-distance point cloud data to be equal to or lower than the second threshold value, where the first threshold value is higher than the second threshold value. Or (B) the coordinate points of the short-distance point cloud data so that the coordinate point density in the short-distance point cloud data is less than or equal to the first threshold value while maintaining the number of coordinate points of the long-distance point cloud data It is a process to reduce the number of.

According to the present technology described above, the acquired point cloud data is reduced in data amount because the number of coordinate points is reduced by the data amount reduction processing. Therefore, it is possible to increase the efficiency or speed of integration between the point cloud data.
Further, in the data amount reduction process, the coordinate point density of the point cloud data is set so that the coordinate point density of the short-distance point cloud data is equal to or lower than the first threshold value and the coordinate point density of the long-distance point cloud data is equal to or lower than the second threshold value. The number is reduced and the first threshold is higher than the second threshold. Therefore, the coordinate point density of the short distance point cloud data after the data amount reduction process can be set to a value suitable for generating a highly accurate map.
Therefore, even if the data amount reduction process is performed to increase the efficiency or speed of integration of the point cloud data by the data amount reduction process, it is possible to obtain point cloud data suitable for generating a highly accurate map.

  The number of coordinate points of the short-distance point cloud data and the long-distance point cloud data is set so that the density of coordinate points in the short-distance point cloud data is equal to or less than the first threshold by the data amount reduction process. Even if it reduces, said effect is acquired. That is, since the density of the coordinate points in the short-distance point cloud data tends to be high, if the density of the coordinate points in the short-distance point cloud data is reduced to an appropriate threshold (first threshold) or less, the coordinates of the long-distance point cloud data It is not necessary to reduce the number of points. This also makes it possible to obtain point cloud data suitable for generating a map with high accuracy while performing a data amount reduction process in order to increase the efficiency or speed of integration between the point cloud data.

1 illustrates a mobile robot system according to an embodiment of the present technology. It is a block diagram which shows the structure of the mobile robot by embodiment of this technique. It is a flowchart which shows the point cloud data processing method by embodiment of this technique. It is explanatory drawing of integration of point cloud data by embodiment of this technique. It is another explanatory drawing of integration of point cloud data by the embodiment of this art. It is another explanatory drawing of integration of point cloud data by the embodiment of this art. It is explanatory drawing of attitude | position estimation of the mobile robot based on the long distance point cloud data by embodiment of this technique.

  An embodiment of the present technology will be described based on the drawings. In addition, the same code | symbol is attached | subjected to the common part in each figure, and the overlapping description is abbreviate | omitted.

  FIG. 1 shows a mobile robot system 1 according to an embodiment of the present technology. The mobile robot system 1 includes a plurality of mobile robots 10. Each mobile robot 10 may be a vehicle that has wheels and travels on the ground surface by being rotationally driven while the wheels are in contact with the ground. Instead, each mobile robot 10 may be another mobile device such as a travel device that travels on the ground by a crawler or a robot that has a leg to walk.

  The mobile robot 10 is provided with an environment measurement sensor 3. The environment measurement sensor 3 uses each measured point on the object surface in the measurement range as viewed from the measurement position, which is the position of the mobile robot 10, as a coordinate point, and the position (coordinate) of each coordinate point as a sensor coordinate as a set coordinate system. Acquires point cloud data represented by a system. The sensor coordinate system is a coordinate system fixed to the mobile robot 10. In the present embodiment, the sensor coordinate system is a three-dimensional coordinate system, and the origin of the sensor coordinate system is the position of the mobile robot 10 when the point cloud data is acquired by the environmental measurement sensor 3 (hereinafter simply referred to as observation). It becomes. In the example of FIG. 1, the measurement range is around the mobile robot 10, and in FIG. 1, each small white circle indicates a measurement point (coordinate point).

  The environmental measurement sensor 3 may be a laser scanner that scans a measurement range with a laser beam (for example, a pulsed laser beam) and measures a three-dimensional position of each coordinate point on an object surface within the measurement range. The laser scanner may be a device called, for example, LiDAR (Light Detection and Ranging) or LRF (Laser Range Finder). In this case, the laser scanner emits laser light in each emission direction from the installation location in the mobile robot 10 to the measurement range. Thus, the laser scanner detects the position of the coordinate point that is the reflection point of the laser beam based on the reflected light of the laser beam for each emission direction.

  However, the environment measurement sensor 3 may be any sensor that can measure the position of each coordinate point in an object existing within the measurement range. For example, the environment measurement sensor 3 is a camera that acquires an image (for example, a stereo image) of a measurement range, and an image that acquires the three-dimensional position of each coordinate point on the surface of an object existing in the measurement range by processing this image. You may have a process part.

  The point cloud data acquired by the environment measurement sensor 3 includes short-distance point cloud data and long-distance point cloud data. The long-distance point cloud data is data of a spatial region that is farther from the measurement position than the short-distance point cloud data. In other words, in the sensor coordinate system, the short-distance point cloud data indicates the position (coordinates) of each coordinate point in the short-distance area on the measurement position side, and the long-distance point cloud data is obtained from the measurement position rather than the short-distance area. Indicates the position (coordinates) of each coordinate point in the far distance area. For example, a spatial region in which the distance from the measurement position where the point cloud data is measured is within a set distance is a short-range region, and a spatial region in which the distance from the measurement position exceeds the set distance is a long-distance region. Here, the set distance is, for example, a value from 20 m to 50 m, a value from 50 m to 75 m, a value from 75 m to 100 m, a value from 100 m to 150 m, or a value from 150 m to 200 m. It is not limited.

  In the point cloud data, the density of coordinate points tends to increase as the position is closer to the measurement position (mobile robot 10). That is, when the same object is measured by the environmental measurement sensor 3 from the first and second measurement positions from the same direction, respectively, and the first measurement position is closer to the object than the second measurement position, The density of coordinate points on the object surface in the point cloud data obtained by measurement from the first measurement position is higher than the density of coordinate points on the object surface in the point cloud data obtained by measurement from the second measurement position. high. Therefore, the density of coordinate points in the short distance point group data is usually higher than the density of coordinate points in the long distance point group data.

  FIG. 2 is a block diagram showing the configuration of the mobile robot 10. In FIG. 2, a plurality of components in one mobile robot 10 are shown. Hereinafter, the configuration of the mobile robot 10 will be described, but each mobile robot 10 may have the same configuration. The mobile robot 10 includes the environment measurement sensor 3, the point cloud data processing device 20 according to the embodiment of the present technology, and the communication unit 7.

  The point cloud data processing device 20 includes a data amount reduction unit 5, a map storage unit 9, an estimation unit 11, and a map generation unit 13.

  The data amount reduction unit 5 reduces the number of coordinate points included in the point cloud data by performing a data amount reduction process on the point cloud data. In the present embodiment, the data amount reducing unit 5 reduces the data amount so that the coordinate point density of the short-distance point group data is less than or equal to the first threshold value and the coordinate point density of the long-distance point group data is less than or equal to the second threshold value. Process. The first threshold is higher than the second threshold. The first threshold value may be a value that is twice or more of the second threshold value, but is not limited to this, for example, a value that is 1.5 times or more of the second threshold value, or a value that is three times or more of the second threshold value. It may be. Although it is self-evident, in the data amount reduction processing in which the coordinate point density of the point cloud data is set to a threshold value (first or second threshold value) or less, when this threshold value is increased N times, the distance between the coordinate points becomes 1 / N. As a result, a data amount reduction effect is reduced as the threshold value is increased.

  The data amount reduction unit 5 inputs the short-distance point cloud data and the long-distance point cloud data after the data amount reduction processing to the communication unit 7, and inputs the long-distance point cloud data after the data amount reduction processing to the estimation unit 11. The short-distance point cloud data and the long-distance point cloud data after the data amount reduction processing are input to the map generation unit 13. In the present embodiment, the short-distance point cloud data after the data amount reduction process is not input to the estimation unit 11. Hereinafter, when “point cloud data” is simply described, “point cloud data” means both short-distance point cloud data and long-distance point cloud data.

  The communication unit 7 transmits the input point cloud data to another mobile robot 10. Further, the communication unit 7 receives point cloud data transmitted from another mobile robot 10. In this case, the communication unit 7 inputs the received point cloud data to the map generation unit 13. In the present embodiment, the communication unit 7 transmits and receives point cloud data by wireless communication.

  The map storage unit 9 stores map data representing the position (coordinates) of each coordinate point on the object surface in the map coordinate system. These coordinate points may include coordinate points registered in advance as points on the surface of a known object (for example, a building). The map coordinate system may be a coordinate system fixed to the ground surface, for example. Map data in the map storage unit 9 is newly generated by the map generation unit 13 as described later.

<Estimation of posture and position based on long-distance point cloud data>
The estimation unit 11 estimates the posture and position of the mobile robot 10 based on the long-distance point cloud data after the data amount reduction process. In the present embodiment, the estimation unit 11 performs processing as follows. The estimation unit 11 reads map data from the map storage unit 9. Next, the estimation unit 11 aligns the input long-distance point group data by translating and rotating the coordinate points of the read map data. Thereafter, the estimation unit 11 obtains the posture and position of the mobile robot 10 with respect to the long-distance point cloud data aligned in the map data.

The long-distance point cloud data input to the estimation unit 11 is represented by a sensor coordinate system fixed to the mobile robot 10 at the time of observation. Therefore, when the estimation unit 11 aligns the long-distance point group data with respect to each coordinate point of the map data, both the following (1) and (2) are integrated with each other in the map data (that is, the map coordinate system). Translate and rotate in.
(1) Long-distance point cloud data (that is, each coordinate point in the sensor coordinate system) input to the estimation unit 11
(2) An index indicating the posture and position of the mobile robot 10 in the sensor coordinate system at the time of observation (for example, the coordinate axis and origin of the sensor coordinate system)
In the map data, the estimation unit 11 estimates the posture and position of the mobile robot from the index thus translated and rotated.

  The above-described alignment by the estimation unit 11 may be performed using, for example, an alignment technique of an ICP (Iterative Closest Point) method or an NDT (Normal Distributions Transform) method.

<Integration processing of point cloud data: map generation>
The map generation unit 13 generates new map data by integrating the input point cloud data with the latest map data in the map storage unit 9. The map generation unit 13 stores the new map data in the map storage unit 9 as the latest map data. The integration by the map generation unit 13 may be performed using the alignment technique described above. With this alignment technique, the map generation unit 13 aligns the input point group data by translating and rotating the coordinate points in the map data in the map coordinate system. When the positioning is completed, the map generation unit 13 integrates the point cloud data after the positioning into map data (that is, point cloud data as each coordinate point in the map data). That is, each coordinate point of the point cloud data after alignment is added to the map data. As a result, the number of coordinate points in the map data increases by the number of added coordinate points. At this time, when a plurality of coordinate points are present in the unit volume of the map data, the plurality of coordinate points are replaced with one representative coordinate point, and one representative coordinate point is included in the unit volume. May be present. Or you may maintain the state which added each coordinate point of the point cloud data after alignment to map data, without performing such replacement.

  The mobile robot 10 includes a control device 15. The control device 15 controls the movement of the mobile robot 10 based on the posture and position of the mobile robot 10 estimated by the estimation unit 11 and the latest map data stored in the map storage unit 9. For example, a movement route that is directed to a preset final target point or waypoint and that does not interfere with an obstacle indicated by each coordinate point of the map data is generated. The control device 15 controls the mobile robot 10 so that the mobile robot 10 moves along the generated movement route. In this control, for example, the control device 15 controls operations of a plurality of actuators that respectively operate an accelerator, a brake, a steering, a transmission, and the like of the vehicle as the mobile robot 10.

(Process flow)
FIG. 3 is a flowchart illustrating a point cloud data processing method according to an embodiment of the present technology. The point cloud data processing method includes steps S1 to S8 and is performed using the above-described point cloud data processing device 20. Steps S1 to S8 may be performed when the mobile robot 10 is moving.

  In step S <b> 1, the environment measurement sensor 3 acquires the above point cloud data by performing measurement from the measurement position that is the current position of the mobile robot 10. In step S <b> 1, the environment measurement sensor 3 inputs the acquired point cloud data to the data amount reduction unit 5.

  In step S2, the data amount reduction unit 5 performs the above-described data amount reduction process on the point cloud data input in step S1. This data amount reduction process may be a filter process. The filter process includes, for example, step S21 and step S22.

  In step S21, the data amount reduction unit 5 divides the three-dimensional space region of the short-distance point cloud data (that is, the above-mentioned short-distance region) into cells having the same first volume (for example, a cube or a rectangular parallelepiped) in the sensor coordinate system. . In step S21, the data amount reduction unit 5 divides the three-dimensional space region (that is, the above-described long-distance region) of the long-distance point group data into cells having the same second volume (for example, a cube or a rectangular parallelepiped). The first volume is smaller than the second volume.

  In step S22, the number of coordinate points existing in each cell divided in step S21 is set to 1 or less. That is, for each cell delimited at step S21, when there are a plurality of coordinate points in the cell, the plurality of coordinate points are replaced with one representative coordinate point, and the representative coordinate point is set in the cell. .

The representative coordinate point setting method in step S22 may be, for example, the following (A) or (B).
(A) For a cell in which a plurality of coordinate points exist, the coordinate value of the representative coordinate point related to each coordinate axis of the sensor coordinate system is determined as the average value of the coordinate values related to the coordinate axis of the plurality of coordinate points. The representative coordinate point of the coordinate value thus determined is set in the cell. In this case, for a cell including only one coordinate point, this coordinate point is set as the representative coordinate point of the cell, and no representative coordinate point is set for a cell not including the coordinate point.
(B) The representative coordinate point of each cell including one or more coordinate points is set at the center of the cell. In this case, no representative coordinate point is set in a cell that does not include a coordinate point.

  By step S21 and step S22, the number of coordinate points existing in each cell of the three-dimensional space area of the sensor coordinate system is 1 or less. Since the reciprocal of the first volume corresponds to the first threshold value and the reciprocal of the second volume corresponds to the second threshold value, the coordinate point density of the short-distance point cloud data becomes equal to or lower than the first threshold value in steps S21 and S22. The coordinate point density of the long-distance point group data is equal to or less than the second threshold value. The filter process described above may be a process using a voxel grid filter, for example. However, in this case, in the above-described step S21, the volume of each voxel as the above-described cell becomes the first volume in the short-distance point group data and the second volume in the long-distance point group data as described above.

  Step S3 and step S4 are performed after step S2. Step S3 and step S4 may be performed in parallel.

  In step S <b> 3, the point cloud data subjected to the data amount reduction process in step S <b> 2 is transmitted to another mobile robot 10 by the communication unit 7. At this time, the communication unit 7 may transmit the long-distance point cloud data and the short-distance point cloud data after the data amount reduction processing to other mobile robots 10 at different points in time.

  In step S <b> 4, the data amount reduction unit 5 inputs the long-distance point cloud data after the data amount reduction processing to the estimation unit 11, and inputs the point cloud data after the data amount reduction processing to the map generation unit 13. In the present embodiment, at this time, the data amount reduction unit 5 first inputs the long-distance point cloud data to the estimation unit 11 and then inputs the point cloud data to the map generation unit 13. Thereby, step S5 can be started early.

  In step S5, the estimation unit 11 estimates the posture and position of the mobile robot 10 in the map data based on the long distance point cloud data input in step S4. That is, the estimation unit 11 reads the latest map data from the map storage unit 9, and inputs the long-distance point group data input to each coordinate point of the map data together with the above-described index representing the posture and position of the mobile robot 10. Rotate and translate to align. The estimation unit 11 estimates the posture and position of the mobile robot 10 from the index after the alignment in the map data.

  In step S6, the map generation unit 13 integrates the point cloud data input in step S4 with the latest map data stored in the map storage unit 9 as described above, thereby obtaining new map data. Generate. In step S6, the map generation unit 13 stores the new map data in the map storage unit 9 as the latest map data.

  On the other hand, when the communication unit 7 receives point cloud data from another mobile robot 10, the communication unit 7 inputs the received point cloud data to the map generation unit 13 (this process is referred to as step S7). Also in this case, step S6 is performed. That is, in step S6, the map generation unit 13 generates new map data by integrating the point cloud data input in step S7 with the latest map data in the map storage unit 9 as described above. . The map generation unit 13 stores the new map data in the map storage unit 9 as the latest map data.

  In step S <b> 8, the control device 15 controls the movement of the mobile robot 10 based on the position and orientation of the mobile robot 10 estimated in step S <b> 5 and the latest map data stored in the map storage unit 9. . For example, the following control is performed. The control device 15 searches and generates a movement route toward a preset final target point or waypoint in the map data. At this time, the control device 15 generates a movement route that does not interfere with an object recognized as an obstacle by each coordinate point in the map data. Next, the control device 15 controls the mobile robot 10 so that the mobile robot 10 moves on the generated movement route.

<Repeat processing>
Steps S1 to S8 described above are repeated. For example, step S1 may be performed at a preset measurement cycle. In this case, steps S2 to S4 may be performed every time step S1 is performed.
On the other hand, every time point cloud data is input to the map generator 13 in step S4 or step S7, step S6 may be performed on this point cloud data.

  Step S5 may be performed at a constant estimation cycle. This estimation cycle may be shorter than the measurement cycle in which step S1 is performed. Hereinafter, the execution time of step S5 that arrives at the estimation cycle is referred to as an estimation time. If no long-distance point cloud data is input to the estimation unit 11 during the period from the previous estimation time point to the current estimation time point, the posture and position are estimated as in the following (a) and (b). Good.

(A) The estimation unit 11 estimates the posture of the mobile robot 10 based on information from the posture sensor 17. The posture sensor 17 measures the amount of posture change of the mobile robot 10 between the previous estimation time and the current estimation time. The estimation unit 11 determines the posture of the mobile robot 10 in the latest map data in the map storage unit 9 at the current estimation time based on the posture change amount and the posture of the mobile robot 10 obtained at the previous estimation time. presume.

  The attitude sensor 17 may be configured using, for example, a gyro sensor, but is not limited to this. For example, the attitude sensor 17 may have a configuration for estimating the attitude of the mobile robot 10 based on radio waves from a positioning satellite in the satellite navigation system, or may be a combination of this configuration and a gyro sensor. Good.

(B) The estimation unit 11 estimates the position of the mobile robot 10 based on information from the speed sensor 19 and the posture sensor 17. The speed sensor 19 measures the moving speed of the mobile robot 10 at each point in the period from the previous estimated point to the current estimated point. The posture sensor 17 measures the posture of the mobile robot 10 at each point in the period from the previous estimated point to the current estimated point. The estimation unit 11 estimates the position of the mobile robot 10 at the current estimation time point based on the measured movement speed and posture at each time point and the position of the mobile robot 10 obtained at the previous estimation time point.

  The speed sensor 19 may measure, for example, the rotational speed of a traveling tire of a vehicle that is the mobile robot 10 and estimate the travel speed of the mobile robot 10 from this rotational speed, but is not limited thereto. . For example, the speed sensor 19 may have a configuration for estimating the moving speed of the mobile robot 10 based on radio waves from a positioning satellite in the satellite navigation system, or a configuration for measuring this rotational speed. It may be a combination.

  In addition, the estimation part 11 may estimate the attitude | position and position of the mobile robot 10 using a Kalman filter with the above-mentioned estimation period.

  The data amount reduction unit 5, the estimation unit 11, and the map generation unit 13 may be configured by a computer, an electronic circuit, or a combination thereof that realizes the above-described function (processing).

(Effect of embodiment)
According to the embodiment of the present technology described above, since the number of coordinate points of the acquired point cloud data is reduced by the data amount reduction process, the data amount is reduced. Therefore, it is possible to increase the efficiency or speed of data communication between the mobile robots 10 and integration of point cloud data. In the data amount reduction processing, the number of coordinate points of the point cloud data is set so that the coordinate point density of the short-distance point cloud data is equal to or lower than the first threshold value and the coordinate point density of the long-distance point cloud data is equal to or lower than the second threshold value. The first threshold is higher than the second threshold. Therefore, the coordinate point density of the short distance point cloud data after the data amount reduction process can be set to a value suitable for generating a highly accurate map.
Therefore, even if the data amount reduction process is performed for the efficiency or speedup of the integration of the point cloud data by the data amount reduction process, short-distance point cloud data suitable for generating a highly accurate map can be obtained.

  4-6 is explanatory drawing of such an effect. As shown in the plan views of FIGS. 4 (A) and 4 (B), FIGS. 4 (A) and 4 (B) assume the vicinity of the crossroads, and obstacles indicated by alternate long and short dash lines define the crossroads. 4 to 6, reference numerals 10 </ b> A and 10 </ b> B denote separate mobile robots 10. As shown in FIGS. 4A and 4B, it is assumed that the two mobile robots 10A and 10B have acquired point cloud data by the environment measurement sensor 3, respectively. The point cloud data acquired by the environment measurement sensor 3 of one mobile robot 10A includes the measurement points of the white circles shown in FIG. 4A as coordinate points, and is acquired by the environment measurement sensor 3 of the other mobile robot 10B. The point cloud data includes the measurement points of each white circle shown in FIG. 4B as coordinate points. In FIGS. 5 and 6, white circles indicate coordinate points. The broken line in FIG. 4A shows the boundary between the short distance area and the long distance area related to the mobile robot 10A, and the broken line in FIG. 4B shows the boundary between the short distance area and the long distance area related to the mobile robot 10B. This is also shown in FIG.

  FIG. 5A shows point cloud data obtained by performing the above-described data amount reduction processing on the point cloud data of FIG. FIG. 5B shows point cloud data obtained by performing the above-described data amount reduction processing on the point cloud data of FIG. As shown in FIGS. 5A and 5B, the density of the coordinate points of the short-distance point group data can be lowered within a range suitable for generating a highly accurate map by the data amount reduction processing. Thereby, for example, the integration of the point cloud data in FIG. 5A and the point cloud data in FIG. 5B can be made efficient or faster. FIG. 6 shows the result of integrating the point cloud data of FIG. 5A and the point cloud data of FIG. As a result, in FIG. 6, a highly accurate map can be generated in the short distance region of FIGS. 5A and 5B, since the number of coordinate points of the long-distance point cloud data is also reduced, the posture and position estimation process (step S5) of the mobile robot 10 using the long-distance point cloud data. ) Is also more efficient or faster.

Since the estimation unit 11 estimates the posture of the mobile robot 10 based on the long-distance point cloud data, the posture is estimated with good accuracy. FIG. 7 is an explanatory diagram of this effect. FIG. 7A shows a state where the object is located close to the mobile robot 10. FIG. 7B shows a state where the same object is located far from the mobile robot 10. The orientation of the mobile robot 10 relative to the measurement point on the object surface has a blur width of an angle α in FIG. 7A and a blur width of an angle β in FIG. 7B. Since α> β, the posture of the mobile robot 10 is obtained with good accuracy by obtaining the posture of the mobile robot 10 with respect to the measurement point (coordinate point) on the surface of the distant object.
Further, the density of the long-distance point cloud data is already low from the time point when the above step S1 is performed. In this respect as well, the time required for the posture and position estimation processing based on the long-distance point cloud data can be shortened.

  The present technology is not limited to the above-described embodiment, and various changes can be made within the scope of the technical idea of the present technology. For example, any one of the following modification examples 1 to 6 may be employed alone, or two or more of modification examples 1 to 6 may be arbitrarily combined and employed. In this case, the points not described below are the same as described above.

(Modification 1)
The data amount reduction unit 5 separates the input point cloud data into short-distance point cloud data and long-distance point cloud data in the data amount reduction process in step S2 described above, and coordinates points of the short-distance point cloud data You may perform the process which makes a density a 1st threshold value, and performs the process which makes the coordinate point density of long-distance point cloud data a 2nd threshold value. In this case, in step S3, the communication unit 7 may transmit the short-distance point cloud data and the long-distance point cloud data separately.

(Modification 2)
In step S <b> 4, the data amount reduction unit 5 may input only the short-distance point group data out of the short-distance point group data and the long-distance point group data after the data amount reduction processing to the map generation unit 13. In this case, in step S <b> 6, the map generation unit 13 generates new map data by integrating the short-distance point cloud data with the latest map data in the map storage unit 9.

  In step S3, the communication unit 7 transmits both the short-distance point cloud data and the long-distance point cloud data after the data amount reduction process in the above description, but the short-distance point cloud data after the data amount reduction process and Only one of the long-distance point cloud data (for example, short-distance point cloud data) may be transmitted to another mobile robot 10. Thereby, in the other mobile robot 10, the said one is received by the communication part 7, and the map production | generation part 13 integrates the said one (for example, short-distance point cloud data) in the newest map data of the map memory | storage part 9. FIG. As a result, new map data is generated.

(Modification 3)
The above-described data amount reduction processing is performed so that the coordinate points of the short-distance point cloud data among the short-distance point cloud data and the long-distance point cloud data are such that the density of the coordinate points in the short-distance point cloud data is equal to or less than the first threshold value. It may be a process of reducing the number. That is, in the data amount reduction process, the number of coordinate points of the long-distance point group data may be maintained without being reduced.

(Modification 4)
Of the plurality of mobile robots 10 constituting the mobile robot system 1, any one of the mobile robots 10 may be replaced with a fixed sensor having no mechanism for moving with respect to the ground surface. This fixed sensor is installed at a set position so as to acquire point cloud data and process the point cloud data. That is, this fixed sensor may be the point cloud data processing device 20 configured to include the environment measurement sensor 3, the data amount reduction unit 5, and the communication unit 7 described above. The point cloud data processing device 20 may not include the map storage unit 9, the estimation unit 11, and the map generation unit 13. In this case, the point cloud data processing device 20 transmits the point cloud data subjected to the data amount reduction processing by the data amount reduction unit 5 from the communication unit 7 to the mobile robot 10 (the mobile robot 10 constituting the mobile robot system 1). Send to. Thereby, in the mobile robot 10, map data is produced | generated by above-mentioned step S6 based on this point cloud data.

(Modification 5)
In the above-described embodiment, the first threshold value is the same value in any part of the short-distance region, and the second threshold value is the same value in any part of the long-distance region. However, the first threshold value may be gradually or stepwise lowered as the distance from the measurement position is increased, and the second threshold value may be gradually or stepwise decreased as the distance from the measurement position is increased.

(Modification 6)
The plurality of mobile robots 10 included in the mobile robot system 1 may not include the point cloud data processing device 20 but may include the mobile robot 10 having the same configuration as that described above. That is, in the mobile robot system 1 according to the present technology, at least one of the mobile robots 10 includes the point cloud data processing device 20.

DESCRIPTION OF SYMBOLS 1 Mobile robot system, 3 Environment measurement sensor, 5 Data amount reduction part, 7 Communication part, 9 Map memory | storage part, 10 Mobile robot, 11 Estimation part, 13 Map generation part, 15 Control apparatus, 17 Posture sensor, 19 Speed sensor, 20 point cloud data processor

Claims (8)

A point cloud data processing device for processing point cloud data acquired by an environmental measurement sensor,
The environmental measurement sensor acquires point cloud data representing the position of each coordinate point in a set coordinate system, with each measured point on the object surface in the measurement range viewed from the measurement position as a coordinate point,
The point cloud data processing device includes:
A data amount reduction unit that reduces the number of coordinate points included in the point cloud data by performing a data amount reduction process on the point cloud data,
The point cloud data includes short-distance point cloud data and long-distance point cloud data. In the set coordinate system, the short-distance point cloud data is a value of each coordinate point in the short-distance region on the measurement position side. Indicating a position, the long-distance point cloud data indicates a position of each coordinate point of the long-distance region that is farther from the measurement position than the short-distance region,
The data amount reduction process is:
(A) A process of setting the coordinate point density of the short-distance point group data to a first threshold value or less and setting the coordinate point density of the long-distance point group data to a second threshold value or less, wherein the first threshold value is the second threshold value. The processing is higher than a threshold value, or (B) while maintaining the number of coordinate points of the long-distance point cloud data, the coordinate point density in the short-distance point cloud data is less than or equal to a first threshold value. A point cloud data processing apparatus, which is a process for reducing the number of coordinate points of distance point cloud data. The data amount reduction unit
In the set coordinate system, the short-distance region is divided into cells of a first volume equal to each other, the long-distance region is divided into cells of a second volume equal to each other, and the first volume is smaller than the second volume,
2. The plurality of coordinate points are replaced with one representative coordinate point when there are a plurality of coordinate points in the cell for each of the cells in the short-distance region and the long-distance region. Point cloud data processing device. The point cloud data processing device is provided in a mobile robot,
The point cloud according to claim 1, further comprising an estimation unit that receives the long-distance point cloud data from the data amount reduction unit and estimates the posture and position of the mobile robot based on the long-distance point cloud data. Data processing device.   4. The apparatus according to claim 1, further comprising a map generation unit that receives at least the short-distance point cloud data from the data amount reduction unit and generates map data based on the short-distance point cloud data. Point cloud data processing device. At least the short-distance point cloud data is input from the data amount reduction unit, and includes a map generation unit that generates map data based on the short-distance point cloud data,
The data amount reduction unit repeatedly performs the data amount reduction process on the point cloud data repeatedly acquired by the environmental measurement sensor,
The map generation unit repeatedly generates the map data based on the short distance point cloud data after the data amount reduction process, which is repeatedly input from the data amount reduction unit,
A map storage unit for storing the latest map data generated by the map generation unit;
The estimation unit estimates the posture and position of the mobile robot in the map data based on the latest map data and the long-distance point cloud data stored in the map storage unit. The point cloud data processor described. A mobile robot,
A point cloud data processing device according to any one of claims 1 to 5,
The environmental measurement sensor for acquiring the point cloud data;
A mobile robot comprising a communication unit that transmits one or both of the short-distance point cloud data and the long-distance point cloud data after the data amount reduction processing to another mobile robot. A mobile robot system comprising a plurality of mobile robots,
Each mobile robot
An environmental measurement sensor that obtains point cloud data representing the position of each coordinate point in a set coordinate system, with each measured point on the object surface in the measurement range viewed from the measurement position that is the position of the mobile robot as a coordinate point;
A communication unit that communicates with other mobile robots,
At least one of the mobile robots is
The point cloud data processing device according to any one of claims 1 to 6, wherein the point cloud data acquired by the environmental measurement sensor of the mobile robot is processed.
The communication unit of the mobile robot transmits one or both of the short-distance point cloud data and the long-distance point cloud data after the data amount reduction processing by the point cloud data processing device to the other mobile robot. ,
The other mobile robot includes a map generation unit that generates map data based on one or both of the transmitted short-distance point cloud data and the long-distance point cloud data. A point cloud data processing method for processing point cloud data acquired by an environmental measurement sensor,
When the point measurement data representing the position of each coordinate point in a set coordinate system is acquired by the environmental measurement sensor with each measured point on the object surface in the measurement range viewed from the measurement position as a coordinate point,
By performing a data amount reduction process on the point cloud data, the number of coordinate points included in the point cloud data is reduced,
The point cloud data includes short-distance point cloud data and long-distance point cloud data. In the set coordinate system, the short-distance point cloud data is a value of each coordinate point in the short-distance region on the measurement position side. Indicating a position, the long-distance point cloud data indicates a position of each coordinate point of the long-distance region that is farther from the measurement position than the short-distance region,
The data amount reduction process is:
(A) A process of setting the coordinate point density of the short-distance point group data to a first threshold value or less and setting the coordinate point density of the long-distance point group data to a second threshold value or less, wherein the first threshold value is the second threshold value. The processing is higher than a threshold value, or (B) while maintaining the number of coordinate points of the long-distance point cloud data, the coordinate point density in the short-distance point cloud data is less than or equal to a first threshold value. A point cloud data processing method, which is a process of reducing the number of coordinate points of distance point cloud data.

Images