JP7715538B2 - Location information processing device and location information processing method - Google Patents

Description

The present invention relates to a location information processing device and a location information processing method.

Conventionally, the route traveled by vehicles such as automobiles or trains and the location of the surrounding environment are measured as a three-dimensional point cloud using a measurement vehicle. Based on the measured three-dimensional point cloud, a high-precision map including, for example, roads and the surrounding environment is created.

Figure 1(A) shows a measurement vehicle, and Figure 1(B) shows how measurements are being taken using the measurement vehicle.

The measurement vehicle 100 has a GNSS device 101, an IMU device 102, an odometry device 103, a laser scanner 104, and a camera 105. The GNSS device 101 receives radio waves from GNSS positioning satellites and estimates the position and attitude of the measurement vehicle 100. The IMU device 102 estimates the position and attitude of the measurement vehicle 100 based on the outputs of an accelerometer and a gyroscope arranged on the measurement vehicle 100. The odometry device 103 estimates the position and attitude of the measurement vehicle 100 based on the speed of the measurement vehicle 100 and the outputs of the gyroscope. The position and attitude of the measurement vehicle 100 are estimated based on the measurement results of the GNSS device 101, the IMU device 102, and the odometry device 103. The position and attitude of the measurement vehicle 100 are expressed, for example, in a world coordinate system with a predetermined position as the origin.

While traveling along a predetermined route, the measurement vehicle 100 uses a laser scanner 104 to scan the area around the measurement vehicle 100 with a laser at measurement times having a predetermined cycle, measuring the relative distance and orientation of multiple positions representing the route and surrounding environment relative to the measurement vehicle 100. The measurement vehicle 100 also photographs the route and surrounding environment using a camera 105.

In addition, the measurement vehicle 100 uses the GNSS device 101, the IMU device 102, and the odometry device 103 to measure the position and orientation of the measurement vehicle 100 at the time of laser scanning.

The position of the travel route and surrounding environment is expressed in the world coordinate system based on the relative distance and direction to the measurement vehicle 100 measured by laser scanning, and the position and attitude information of the measurement vehicle 100 during laser scanning.

In this way, a three-dimensional point cloud representing the position of the travel path and surrounding environment is obtained along the travel path of the measurement vehicle 100.

Here, radio waves from GNSS positioning satellites may be blocked from the measurement vehicle 100 by tunnels, high-rise buildings, etc. If radio waves from the positioning satellites are blocked from the measurement vehicle 100, the accuracy of the position and attitude of the measurement vehicle 100 may decrease.

The measurement accuracy of the travel path and the surrounding environment of the measurement vehicle 100 measured by laser scanning is thought to be high, but as the accuracy of the position and orientation of the measurement vehicle 100 decreases during laser scanning, the accuracy of the position of the 3D point cloud representing the position of the surrounding environment of the travel path decreases. Therefore, it has been proposed to correct the position of the measured 3D point cloud (for example, see Patent Document 1).

Japanese Patent Application Laid-Open No. 2020-015419

Because the driving path and surrounding environment can change, the same driving path and surrounding environment positions may be measured at different times. As a result, two 3D point clouds are obtained.

However, as mentioned above, due mainly to the accuracy of the position and attitude information of the measurement vehicle 100, the positions representing the driving route and the surrounding environment may differ between two 3D point clouds measured for the same driving route.

To investigate changes in the driving route and surrounding environment, it is necessary to correct the 3D point clouds so that the two 3D point clouds can be compared.

The present specification therefore aims to provide a position information processing device that corrects three-dimensional point clouds so that two separately measured three-dimensional point clouds can be compared.

According to one embodiment, a position information processing device is provided. The position information processing device is configured to: determine whether a first reference line representing a predetermined linear environment is set for a first three-dimensional point cloud representing the position of the environment surrounding a first travel path of a first vehicle, the first reference line being set for a second three-dimensional point cloud representing the position of the environment surrounding a second travel path that includes at least a portion of the first travel path, the second reference line being set for a second three-dimensional point cloud representing the position of the environment surrounding a second travel path that includes at least a portion of the first travel path, and the second reference line representing the predetermined linear environment is set for a second three-dimensional point cloud representing the position of the environment surrounding a second travel path that includes at least a portion of the first travel path, the second reference line being set for a second three-dimensional point cloud representing the position of the environment surrounding a second travel path that includes at least a portion of the first travel path, and the second reference line being set for a second three-dimensional point cloud representing the position of the environment surrounding a second travel path that includes at least a portion of the first travel path, is greater than a predetermined first reference distance; a section determination unit that determines a separation section in which a distance between a start position of the first reference line and a second reference line that represents a predetermined linear environment corresponding to the first reference line is greater than a first reference distance; an association unit that associates each of a plurality of first reference points that form the separation section of the first reference line with a point included in the first three-dimensional point cloud; an association unit that associates each of a plurality of first reference points that form the separation section of the first reference line with a second reference point that form the separation section of the second reference line; a first correction unit that calculates a difference in position between two reference points and corrects the position of a point included in a first three-dimensional point cloud associated with the first reference point using the difference in position; a straight line generation unit that generates a reference straight line by arranging each of a plurality of first reference points that form a separation section of the first reference line so that the first reference points extend in a predetermined direction from a starting point such that the distance between adjacent first reference points is maintained, and that arranging each of a plurality of second reference points that form a separation section of the second reference line on the reference straight line from the starting point such that the distance between adjacent second reference points is maintained; and a second correction unit that, for each of a plurality of first reference points forming a separation section of one reference line, determines the difference in position between the first reference point and a point on the corresponding reference line, and uses the position difference to correct the position of a point included in a first three-dimensional point cloud associated with the first reference point; and, for each of a plurality of second reference points forming a separation section of a second reference line, determines the difference in position between the second reference point and a point on the corresponding reference line, and uses the position difference to correct the position of a point included in a second three-dimensional point cloud associated with the second reference point.

According to another embodiment, a position information processing method is provided. This position information processing method includes: determining whether a distance between a first reference line representing a predetermined linear environment set for a first three-dimensional point cloud representing the position of the environment surrounding a first travel path of a first vehicle, the first reference line being set for a second three-dimensional point cloud representing the position of the environment surrounding a second travel path that includes at least a portion of the same path as the first travel path, the second reference line being set for a second three-dimensional point cloud representing the position of the environment surrounding a second travel path that includes at least a portion of the same path as the first travel path, is greater than a predetermined first reference distance. a start position where the distance between the first reference line and a second reference line representing a corresponding predetermined linear environment is greater than a first reference distance; each of the first reference points forming the separation section of the first reference line is associated with a point included in the first three-dimensional point cloud; each of the first reference points forming the separation section of the first reference line is associated with a second reference point forming the separation section of the second reference line; and each of the first reference points forming the separation section of the first reference line is associated with the first reference point. the method further comprises: generating a reference line by arranging each of a plurality of first reference points forming a separation section of the first reference line so as to extend in a predetermined direction from a starting point such that the distance between adjacent first reference points is maintained; and arranging each of a plurality of second reference points forming a separation section of the second reference line on the reference line from a starting point such that the distance between adjacent second reference points is maintained; for each of the plurality of first reference points forming the separation section of the first reference line, calculating a difference in position between the first reference point and a point on the reference line corresponding to the first reference point; and correcting the position of the point included in the first three-dimensional point cloud associated with the first reference point using the difference in position; and for each of a plurality of second reference points forming a separation section of the second reference line, calculating a difference in position between the second reference point and a point on the reference line corresponding to the second reference point;

The position information processing device and position information processing method disclosed in this specification described above can correct three-dimensional point clouds so that two separately measured three-dimensional point clouds can be compared.

FIG. 1A is a diagram showing a measurement vehicle, and FIG. 1B is a diagram showing a state in which measurement is performed using the measurement vehicle. 1 is a schematic configuration diagram of a position information processing device according to an embodiment of the present invention; 10 is an operational flowchart relating to position information processing of the position information processing device. FIG. 2 is a diagram showing a first reference line and a second reference line. FIG. 10 is a diagram illustrating an association process. 10A to 10C are diagrams illustrating the association process. FIG. 10 is a diagram illustrating a line generation process. 10A and 10B are diagrams illustrating a process of correcting the position of a first three-dimensional point cloud. FIG. 1A is a diagram showing a first projected image, and FIG. 1B is a diagram showing a second projected image. FIG. 1A is a diagram showing a first linear pattern, and FIG. 1B is a diagram showing a second linear pattern. FIG. 10A is a diagram showing a first cross-sectional image, and FIG. 10B is a diagram showing a second cross-sectional image. 10A is a diagram showing a third linear pattern, and FIG. 10B is a diagram showing a fourth linear pattern. (A) shows a diagram in which the first 3D point cloud whose position has been corrected is superimposed on the second 3D point cloud, and (B) shows a diagram in which the first 3D point cloud before correction is superimposed on the second 3D point cloud.

Below, a preferred embodiment of the location information processing device disclosed in this specification will be described with reference to the drawings.

Figure 2 is a schematic diagram of the position information processing device 10 of this embodiment. The position information processing device 10 corrects one of two 3D point clouds measured separately so that they can be compared. The two 3D point clouds represent the driving route and the position of the surrounding environment measured by a measurement vehicle such as that shown in Figure 1 on a driving route that includes at least a partial identical route. The two 3D point clouds may have been measured by different measurement vehicles, or may have been measured at different times by the same measurement vehicle.

The position information processing device 10 has a communication interface (IF) 21, a user interface (UI) 22, a memory 23, and a processor 24. The communication IF 21, UI 22, memory 23, and processor 24 are connected via a communication line 25.

The communication IF 21 has an interface circuit for connecting the position information processing device 10 to a network (not shown). Data of the two 3D point clouds is input to the position information processing device 10, for example, from the measurement vehicle 100 via the network.

The UI 22 receives user operations, generates operation signals, and outputs the processing results of the position information processing device 10. The UI 22 has an input device, such as a keyboard, mouse, or touch panel, for inputting user operations. The UI 22 has a display device, such as an LCD display or touch panel, as an output device for outputting the processing results.

Memory 23 is an example of a storage unit and includes, for example, volatile semiconductor memory and non-volatile semiconductor memory. Memory 23 stores computer programs that allow processor 24 to execute location information processing, as well as data generated during location information processing.

The memory 23 also stores first three-dimensional point cloud data representing the position of the environment surrounding the first travel route of the first measurement vehicle, collected by scanning with a laser scanner mounted on the first measurement vehicle, and second three-dimensional point cloud data representing the position of the environment surrounding a second travel route that includes at least a partial path that is the same as the first travel route, collected by scanning with a laser scanner mounted on the second measurement vehicle.

The processor 24 includes one or more central processing units (CPUs) and their peripheral circuits. The processor 24 may also include other arithmetic circuits such as a logic arithmetic unit, a numerical arithmetic unit, or a graphics processing unit.

All or some of the functions of the position information processing device 10 are functional modules implemented, for example, by a computer program running on the processor 24. The processor 24 includes a reference line setting unit 31, a section determination unit 32, an association unit 33, a correspondence unit 34, a correction unit 35, and a straight line generation unit 36. Alternatively, the functional modules of the processor 24 may be dedicated arithmetic circuits provided in the processor 24.

Figure 3 is an operational flowchart related to the position information processing of the position information processing device 10. First, the reference line setting unit 31 reads the data of the first and second 3D point clouds from the memory 23, sets a first reference line representing a predetermined linear environment for the first 3D point cloud, and sets a second reference line representing an environment corresponding to the first reference line for the second 3D point cloud (step S101). Figure 4 is a diagram showing the first reference line R1 and the second reference line R2. The first reference line is set based on the position or travel route of the environment measured by the first measurement vehicle, and the second reference line is set based on the position or travel route of the environment measured by the second measurement vehicle.

The reference line setting unit 31 may set a first reference line R1 for the first three-dimensional point cloud using multiple points in the first three-dimensional point cloud, and may set a second reference line R2 using multiple points in the second three-dimensional point cloud.

For example, the reference line setting unit 31 sets a first reference line R1 based on points in a first three-dimensional point cloud included in a predetermined area determined based on position information representing the travel trajectory of the first measurement vehicle that have a laser scanner reflection intensity equal to or greater than a predetermined threshold, and sets a second reference line R2 based on points in a second three-dimensional point cloud included in the predetermined area determined based on position information representing the travel trajectory of the second measurement vehicle that have a laser scanner reflection intensity equal to or greater than the predetermined threshold. If the first and second measurement vehicles are automobiles, the predetermined area can be an area that includes the center line of the road on which the first and second measurement vehicles traveled.

Furthermore, when the first measurement vehicle and the second measurement vehicle are railway vehicles, the reference line setting unit 31 may set a first reference line R1 based on points in the first three-dimensional point cloud that represent the track on which the first measurement vehicle is traveling, and may set a second reference line R2 based on points in the second three-dimensional point cloud that represent the track on which the second measurement vehicle is traveling. The reference line setting unit 31 may be configured to include a classifier that receives input of the three-dimensional point cloud and is trained to detect points that represent the track of the railway vehicle. This classifier receives input of the three-dimensional point cloud and detects points that represent the track of the railway vehicle.

Furthermore, the first and second reference lines may be provided by the user rather than being set by the reference line setting unit 31. For example, if the first and second measurement vehicles are railway vehicles and the first and second travel paths include the same track, the travel path of the first measurement vehicle measured by the first measurement vehicle may be set as the first reference line R1, and the travel path of the second measurement vehicle measured by the second measurement vehicle may be set as the second reference line R2.

Next, the interval determination unit 32 determines a start position where the distance between the first reference line R1 and the second reference line R2 begins to increase beyond a predetermined first reference distance, and a separation interval where the distance between the first reference line R1 and the second reference line R2, which represents the corresponding predetermined linear environment, increases beyond the first reference distance (step S102). If an end position exists where the distance between the first reference line R1 and the second reference line R2 ends to increase beyond the predetermined reference distance, the interval determination unit 32 may determine this end position. In the example shown in FIG. 4, a separation interval D is set between the start position P1 and the end position P2. The first reference distance can be, for example, a distance of 0.1 m to 3 m.

Note that the first reference line R1 and the second reference line R2 extending from the start position may not intersect with each other. In this case, the separation section refers to the section from the start position where the first reference line R1 and the second reference line R2 remain separated and end as an open end.

Next, the association unit 33 associates each of the multiple first reference points forming the separation section D of the first reference line R1 with points included in the first three-dimensional point cloud (step S103). The first reference line R1 is divided into multiple first reference points based on, for example, a distance interval corresponding to the measurement time when the first three-dimensional point cloud was measured. Alternatively, the first reference line R1 may be divided into multiple first reference points at a predetermined distance interval.

Figure 5 is a diagram illustrating the association process. For example, for each of the multiple first reference points forming the separation section D, the association unit 33 associates the first reference point with a point of the first three-dimensional point cloud that is included in the cross section 500 that includes the first reference point and is perpendicular to the first reference line R1.

Furthermore, when the first reference line R1 is set using multiple points in the first three-dimensional point cloud, the associating unit 33 may associate, for each of the multiple first reference points forming the separation section D of the first reference line R1, a point included in the first three-dimensional point cloud whose position was measured at the same time as the first reference point was measured with the first reference point. When the first reference line R1 is set using multiple points in the first three-dimensional point cloud, the first reference point is associated with the point in the first three-dimensional point cloud that represents the first reference point (itself).

Next, the association unit 34 associates each of the multiple first reference points that form the separation section D of the first reference line R1 with the second reference points that form the separation section D of the second reference line R2 (step S104). Figures 6(A) to 6(C) are diagrams that explain the association process.

As shown in FIG. 6(A), the association unit 34 may associate each of the multiple first reference points forming the separation section D of the first reference line R1 with the second reference point that is closest to the first reference point.

Furthermore, as shown in FIG. 6(B), the association unit 34 may associate, for each of the multiple first reference points forming the separation section D, a second reference point such that the ratio l1/L1 of the distance l1 from the start position P1 to the first reference point to the distance L1 of the separation section D on the first reference line R1 matches with the ratio l2/L2 of the distance l2 from the start position P1 to the second reference point to the distance L2 of the separation section D on the second reference line R2.

As shown in Figure 6 (C), once this matching process has been performed, the second reference point on the second reference line R2 that is associated with the first reference point is likely to have a deviation along the second reference line R2 and/or a deviation in the rotation angle around the second reference line R2 as the rotation axis from the actual driving path and the position of the surrounding environment represented by the first reference point. These deviations are corrected in the process described below.

Next, for each of the multiple first reference points forming the separation section D of the first reference line R1, the correction unit 35 calculates the position difference H (see Figure 6 (C)) between this first reference point and the associated second reference point, and uses this position difference H1 to correct the positions of the points included in the first three-dimensional point cloud associated with this first reference point (step S105).

The matrix that transforms a first reference point into a corresponding second reference point (hereinafter also referred to as the first correction matrix) is expressed as a combination of a translation matrix that represents parallel movement and a rotation matrix that represents rotation. Position correction using the position difference H1 can be expressed with this first correction matrix. The correction unit 35 uses the first correction matrix to correct the positions of points included in the first three-dimensional point cloud associated with each of the multiple first reference points that form the separation section D of the first reference line R1. When the first reference line R1 is set using multiple points in the first three-dimensional point cloud, the positions of the points in the first three-dimensional point cloud that form the first reference line R1 are also corrected using the first correction matrix.

Next, the line generation unit 36 generates a reference line R3 by positioning each of the multiple first reference points that form the separation section D of the first reference line R1 so that they extend in a predetermined direction from the start position P1 so that the distance between adjacent first reference points is maintained, and also positions each of the multiple second reference points that form the separation section D of the second reference line R1 on the reference line R3 from the start position P1 so that the distance between adjacent second reference points is maintained (step S106). Points that were associated with the first reference points in the association process described above can be used as the multiple second reference points. An arbitrary direction is given in advance as the predetermined direction.

Figure 7 is a diagram illustrating the line generation process. Each of the multiple first reference points that form the separation section D is placed on a reference line R3 that extends in a predetermined direction from the starting position P1, while maintaining the distance between adjacent first reference points. Similarly, each of the multiple second reference points that form the separation section D is placed on a reference line R3 that starts from the starting position P1, while maintaining the distance between adjacent second reference points.

Next, the correction unit 35 corrects the positions of the points included in the first 3D point cloud and the second 3D point cloud based on the reference line R3 (step S107). Figures 8(A) and 8(B) are diagrams illustrating the process of correcting the positions of the first 3D point cloud. Figure 8(A) shows the first 3D point cloud G1 and the first reference line R1 before their positions are corrected.

For each of the multiple first reference points forming the separation section D of the first reference line R1, the correction unit 35 calculates the positional difference H2 (see Figure 7) between this first reference point and the corresponding point on the reference line R3, and uses this positional difference H2 to correct the position of the point included in the first three-dimensional point cloud associated with this first reference point and corrected by the first correction matrix; and for each of the multiple second reference points forming the separation section D of the second reference line R2, the correction unit 35 calculates the positional difference H3 (see Figure 7) between this second reference point and the corresponding point on the reference line R3, and uses this positional difference H3 to correct the position of the point included in the second three-dimensional point cloud associated with this second reference point R3.

The matrix (hereinafter referred to as the second correction matrix) that transforms the first reference point into a corresponding point on the reference line R3 is expressed as a combination of a translation matrix that represents parallel translation and a rotation matrix that represents rotation. Position correction using the position difference H2 can be expressed with this second correction matrix. The correction unit 35 uses the second correction matrix to correct the positions of points included in the first three-dimensional point cloud that are associated with each of the multiple first reference points that form the separation section D of the first reference line R1 and that have been corrected using the first correction matrix. If the first reference line R1 is set using multiple points in the first three-dimensional point cloud, the positions of the points in the first three-dimensional point cloud that form the first reference line R1 are also corrected using the second correction matrix.

The matrix that transforms the second reference point into a corresponding point on the reference line R3 (hereinafter referred to as the third correction matrix) is expressed as a combination of a translation matrix that represents parallel translation and a rotation matrix that represents rotation. Position correction using the position difference H3 can be expressed with this third correction matrix. The correction unit 35 uses the third correction matrix to correct the position of the point included in the second three-dimensional point cloud associated with each of the multiple second reference points that form the separation section D of the second reference line R2.

Figure 8(B) shows the first three-dimensional point cloud G1 and the reference line R3 whose positions have been corrected using the second correction matrix. The positions of the first three-dimensional point cloud G1 whose positions have been corrected using the second correction matrix can be expressed, for example, using a Cartesian coordinate system in which the reference line R3 is the y-axis direction and the starting position P1 is the origin. As shown in Figure 8(B), the first three-dimensional point cloud G1 is arranged in a straight line around the reference line R3. Although not shown, the second three-dimensional point cloud whose positions have been corrected using the third correction matrix is also arranged in a straight line around the reference line R3.

Next, the correction unit 35 corrects the position of the first three-dimensional point cloud along the reference line R3 (step S108). The correction unit 35 generates a first projected image in which the first three-dimensional point cloud, whose positions have been corrected using the second correction matrix, is projected onto a projection plane including the reference line R3. Figure 9 (A) is a diagram showing the first projected image 900. The pixels of the first projected image 900 have brightness values ranging from 0 to 255, with the brightness of pixels containing points being 255 and the brightness of pixels containing no points being 0. This explanation also applies to the second projected image, which will be described later.

The correction unit 35 also generates a second projection image 901 in which the second three-dimensional point cloud, whose positions have been corrected using the third correction matrix, is projected onto the projection surface including the reference line R3. Figure 9(B) is a diagram showing the second projection image 901. The first projection image 900 and the second projection image 901 are formed by projecting the three-dimensional point cloud onto the same projection surface, but Figure 9(A) shows an image in which only the first three-dimensional point cloud is projected, and Figure 9(B) shows an image in which only the second three-dimensional point cloud is projected.

The correction unit 35 then selects from the first projection image 900 a first line pattern that extends in a direction perpendicular to the reference line R3 and has a length equal to or greater than a predetermined second reference distance. Here, the correction unit 35 preferably selects from the first projection image 900 a first line pattern that is located within a predetermined distance from the reference line R3, extends in a direction perpendicular to the reference line R3, and has a length equal to or greater than the predetermined second reference distance. The second reference distance may be 0.5 m to 5 m. The predetermined distance from the reference line R3 may be 1 m to 5 m. Figure 10(A) is a diagram showing multiple first line patterns E1 to E9.

The correction unit 35 then detects second linear patterns whose shapes match those of the first linear patterns from the second projection image. In the example shown in FIG. 10(A), for each of the first linear patterns E1 to E9, the correction unit 35 detects second linear patterns whose shapes match those of the first linear patterns E1 to E9 from the second projection image 901 within a predetermined range of the first linear patterns E1 to E9. Pixels representing the first and second linear patterns described above can be detected from the projection image using, for example, a Sobel filter. FIG. 10(B) is a diagram showing multiple second linear patterns F1 to F9. In the example shown in FIGS. 10(A) and 10(B), the second linear pattern F1 is detected from the second projection image 901 as a linear pattern whose shape matches that of the first linear pattern E1. Similarly, the second linear patterns F2 to F9 are detected from the second projection image 901 as linear patterns whose shapes match those of the first linear patterns E2 to E9.

The correction unit 35 then calculates the position difference H4 between each of the second line patterns F1 to F9 and the corresponding first line patterns E1 to E9 in the direction of the reference line R3, and uses this position difference H4 to correct the positions of the points on the reference line R3 that correspond to the positions of the first line patterns E1 to E9, as well as the positions of the points included in the first three-dimensional point cloud that are associated with the first reference points that correspond to the points on this reference line R3.

Position correction using the position difference H4 can be expressed by a fourth correction matrix. The fourth correction matrix is expressed as a translation matrix representing parallel movement. The correction unit 35 uses the fourth correction matrix to correct the positions of points on the reference line R3 that correspond to the positions of the first line patterns E1 to E9, and the positions of points included in the first three-dimensional point cloud that are associated with the first reference points that correspond to the points on this reference line R3.

In addition, if the first reference point corresponding to the point on the reference line R3 corresponding to the position of the first straight line pattern is not associated with a point included in the first three-dimensional point cloud, the fourth correction matrix for the first reference point associated with the point included in the first three-dimensional point cloud is obtained by linear interpolation or curved interpolation based on the fourth correction matrix for the position of the point on the reference line R3 corresponding to one or two first straight line pattern positions close to the point on the reference line R3 corresponding to this first reference point.

Next, the correction unit 35 interpolates and corrects the positions of points on the reference line R3 located between two adjacent first line patterns E1-E9, and the positions of points included in the first three-dimensional point cloud associated with the first reference point corresponding to the point on the reference line R3 (step S109). Linear or curvilinear interpolation can be used as the interpolation correction. For points on the reference line R3 that are adjacent to only one of the first line patterns E1-E9, linear or curvilinear interpolation is used by extrapolation. Through this interpolation correction, for example, the positions of points on the reference line R3 located between the first line pattern E1 and the first line pattern E2, and the positions of points included in the first three-dimensional point cloud associated with the first reference point corresponding to the point on the reference line R3, are interpolated and corrected. Similarly, the positions of points on the reference line R3 located between the first linear pattern E2 and the first linear pattern E3, between the first linear pattern E3 and the first linear pattern E4, between the first linear pattern E4 and the first linear pattern E5, between the first linear pattern E5 and the first linear pattern E6, between the first linear pattern E6 and the first linear pattern E7, between the first linear pattern E7 and the first linear pattern E8, and between the first linear pattern E8 and the first linear pattern E9 are interpolated and corrected, as are the positions of points included in the first three-dimensional point cloud associated with the first reference points corresponding to the points on the reference line R3.

Next, the correction unit 35 rotates the positions of the first three-dimensional point cloud around the reference line R3 as an axis to correct the angle (step S110). The correction unit 35 generates a first cross-sectional image representing the first three-dimensional point cloud whose positions have been corrected by the fourth correction matrix, and which includes points on the reference line R3 whose positions have been corrected by the fourth correction matrix and is included in a cross section perpendicular to the reference line R3. Figure 11 (A) is a diagram showing the first cross-sectional image 1100.

The correction unit 35 also generates a second cross-sectional image representing a second three-dimensional point cloud included in the cross-section, the positions of which have been corrected by the third correction matrix. FIG. 11(B) is a diagram showing the second cross-sectional image 1101. The first cross-sectional image 1100 and the second cross-sectional image 1101 represent the same cross-section perpendicular to the reference line R3, but FIG. 11(A) shows an image representing only the first three-dimensional point cloud, while FIG. 11(B) shows an image representing only the second three-dimensional point cloud. The pixels of the first cross-sectional image 1100 and the second cross-sectional image 1101 may have brightness values ranging from 0 to 255, with the brightness of pixels containing points being 255 and the brightness of pixels not containing points being 0.

The correction unit 35 then selects from the first cross-sectional image 1100 a third line pattern that extends in a direction perpendicular to the reference line R3 and has a length equal to or greater than a predetermined third reference distance. Here, the correction unit 35 preferably selects from the first cross-sectional image 1100 a third line pattern that is located within a predetermined distance from the reference line R3, extends in a direction perpendicular to the reference line R3, and has a length equal to or greater than the predetermined third reference distance. The third reference distance may be 0.5 m to 5 m. The predetermined distance from the reference line R3 may be 1 m to 5 m.

Figure 12 (A) is a diagram showing the third linear patterns E10 and E11. The third linear patterns E10 and E11 are perpendicular to each other. Pixels representing the third linear pattern and the fourth linear pattern described below can be detected from the cross-sectional image using, for example, a Sobel filter.

The correction unit 35 then projects the third line pattern onto the second cross-sectional image and rotates the third line pattern around the reference line R3 as the rotation axis to determine the rotation angle at which a fourth line pattern matching the shape of the third line pattern is detected in the second cross-sectional image. The second cross-sectional image onto which the third line pattern is projected may be a cross-sectional image from which pixels representing edges have been extracted using, for example, a Sobel filter. Figure 12(B) shows fourth line patterns F10 and F11. In the example shown in Figures 12(A) and 12(B), the fourth line pattern F10 is detected from the second projection image 901 as a line pattern matching the shape of the third line pattern E10, and the fourth line pattern F11 is detected from the second projection image 901 as a line pattern matching the shape of the third line pattern E11. In the example shown in Figures 12(A) and 12(B), two third line patterns and two corresponding fourth line patterns are detected, but only one third line pattern and one corresponding fourth line pattern may be detected.

The correction unit 35 then uses the determined rotation angle to correct the angle around the reference line R3 of the point included in the first three-dimensional point cloud associated with the first reference point corresponding to the point on the reference line R3 included in the first cross-sectional image 1100. As in the example shown in Figures 12(A) and 12(B), when two third line patterns and two corresponding fourth line patterns are detected, the average value of the two rotation angles may be used for correction.

Position correction using the rotation angle can be expressed by a fifth correction matrix. The fifth correction matrix is expressed as a rotation matrix representing rotation. The correction unit 35 uses the fifth correction matrix to correct the angles, about the reference line R3, of points included in the first three-dimensional point cloud associated with first reference points corresponding to points on the reference line R3 included in the first cross-sectional image 1100. The correction unit 35 calculates a fifth correction matrix for each of the multiple first reference points on the first reference line R1, and uses this fifth correction matrix to correct the angles, about the reference line R3, of points included in the first three-dimensional point cloud associated with the first reference points.

Next, the correction unit 35 performs a correction to return the positions of the first three-dimensional point cloud, the positions of which have been corrected using the fifth correction matrix, to the relationship between the first reference point and the first reference line (step S111). The correction unit 35 corrects the positions of the points included in the first three-dimensional point cloud associated with the first reference point, the positions of which have been corrected using the fifth correction matrix, using the difference H2 between the positions of the first reference point and the point on the reference line R3 corresponding to this first reference point. Specifically, the correction unit 35 corrects the positions of the points included in the first three-dimensional point cloud associated with the first reference point, the positions of which have been corrected using the fifth correction matrix, using a sixth correction matrix representing the inverse matrix of the second correction matrix.

Figure 13(A) shows a diagram in which the first 3D point cloud whose position has been corrected is overlaid on the second 3D point cloud, and Figure 13(B) shows a diagram in which the first 3D point cloud before correction is overlaid on the second 3D point cloud. As shown in Figure 13(B), among the points included in the first 3D point cloud before correction, points that represent the same driving route and surrounding environment as the second 3D point cloud are shifted from the second 3D point cloud. On the other hand, as shown in Figure 13(A), due to the above-mentioned correction, among the points included in the first 3D point cloud, points that represent the same driving route and surrounding environment as the second 3D point cloud are in the same positions.

The position information processing device of this embodiment described above can correct three-dimensional point clouds so that two separately measured three-dimensional point clouds can be compared. Specifically, the first three-dimensional point cloud is converted into a relationship with the reference line, so that the position of the first three-dimensional point cloud can be compared with the second three-dimensional point cloud by correcting the position of the first three-dimensional point cloud along the reference line and correcting the angle of the reference line of the first three-dimensional point cloud around the reference line as the axis of rotation. The position information processing device of this embodiment can achieve faster and more accurate correction processing.

By using a location information processing device to compare two separately measured 3D point clouds, it is possible to update the map database or perform equipment maintenance.

In the present invention, the location information processing device and location information processing method of the above-described embodiments can be modified as appropriate without departing from the spirit of the present invention. Furthermore, the technical scope of the present invention is not limited to these embodiments, but extends to the inventions set forth in the claims and their equivalents.

For example, in the above-described embodiment, the position information processing device further corrects the positions of the first 3D point cloud along the reference line and the angle around the reference line of the first 3D point cloud as the rotation axis using the first 3D point cloud whose positions have been corrected using the second correction matrix and the second 3D point cloud whose positions have been corrected using the third correction matrix. However, the user may manually correct the positions of the first 3D point cloud along the reference line and the angle around the reference line of the first 3D point cloud as the rotation axis.

In the above-described embodiment, the correction unit 35 corrected the angle of the first three-dimensional point cloud by rotating it around the reference line as the rotation axis, but this correction does not need to be performed if the difference in angle is estimated to be small. In this case, the correction unit 35 corrects, using the sixth correction matrix, the positions of the points included in the first three-dimensional point cloud associated with the first reference point whose position has been corrected using the fourth correction matrix.

REFERENCE SIGNS LIST 10 Position information processing device 21 Communication interface 22 User interface 23 Memory 24 Processor 25 Communication line 31 Reference line setting unit 32 Section determination unit 33 Association unit 34 Correspondence unit 35 Correction unit 36 Straight line generation unit

Claims (12)

a section determination unit that determines a start position at which a distance between a first reference line representing a predetermined linear environment and a corresponding second reference line representing the predetermined linear environment, the first reference line being set for a first three-dimensional point cloud that is collected by scanning with a laser scanner mounted on a first vehicle and that represents the position of the environment around a first traveling route of the first vehicle, and a second three-dimensional point cloud that is collected by scanning with a laser scanner mounted on a second vehicle and that represents the position of the environment around a second traveling route that includes at least a portion of the first traveling route, and that is
an associating unit that associates each of a plurality of first reference points that form the separation section of the first reference line with a point included in the first three-dimensional point cloud;
a correspondence unit that associates each of the plurality of first reference points that form the separated section of the first reference line with a second reference point that forms the separated section of the second reference line;
a first correction unit that calculates a difference in position between each of the first reference points forming the separation section of the first reference line and a second reference point associated with the first reference point, and corrects the position of a point included in the first three-dimensional point cloud associated with the first reference point using the difference in position;
a line generating unit that generates a reference line by arranging each of the plurality of first reference points that form the separated section of the first reference line so as to extend in a predetermined direction from the start position as a starting point so as to maintain a distance between adjacent first reference points, and that arranges each of the plurality of second reference points that form the separated section of the second reference line on the reference line so as to maintain a distance between adjacent second reference points, with the start position as a starting point;
a second correction unit that calculates, for each of a plurality of first reference points forming the separated section of the first reference line, a difference in position between the first reference point and a point on the reference line corresponding to the first reference point, and corrects, using the difference in position, a position of a point included in the first three-dimensional point cloud associated with the first reference point corrected by the first correction unit; and that calculates, for each of a plurality of second reference points forming the separated section of the second reference line, a difference in position between the second reference point and a point on the reference line corresponding to the second reference point, and corrects, using the difference in position, a position of a point included in the second three-dimensional point cloud associated with the second reference point;
A position information processing device comprising: The position information processing device of claim 1, further comprising a reference line setting unit that sets the first reference line using a plurality of points in the first three-dimensional point cloud and sets the second reference line using a plurality of points in the second three-dimensional point cloud. The reference line setting unit
setting the first reference line based on points of the first three-dimensional point cloud included in a predetermined area determined based on position information representing a travel trajectory of the first vehicle, the points having a laser scanner reflection intensity equal to or greater than a predetermined threshold value; and
3. The position information processing device according to claim 2, wherein the second reference line is set based on points among the second three-dimensional point cloud included in the predetermined area determined based on position information representing the travel trajectory of the second vehicle, the points having a laser scanner reflection intensity equal to or greater than a predetermined threshold. the first vehicle and the second vehicle are railroad vehicles,
The reference line setting unit
setting the first reference line based on points in the first three-dimensional point cloud that represent a trajectory on which the first vehicle travels; and
The position information processing device according to claim 2 , wherein the second reference line is set based on points in the second three-dimensional point cloud that represent a trajectory on which the second vehicle travels. The associating unit
A position information processing device described in any one of claims 1 to 4, wherein for each of a plurality of first reference points forming the separation section, a point of the first three-dimensional point cloud that includes the first reference point and is included in a cross section perpendicular to the first reference line is associated with the first reference point. The associating unit
A position information processing device described in any one of claims 2 to 4, wherein for each of a plurality of first reference points forming the separation section of the first reference line, a point included in the first three-dimensional point cloud whose position was measured at the same time as the time when the first reference point was measured is associated with the first reference point. The association unit
A position information processing device according to any one of claims 1 to 6, wherein for each of a plurality of first reference points forming the separation section of the first reference line, the second reference point located closest to the first reference point is associated. The association unit
A position information processing device as described in any one of claims 1 to 6, wherein for each of the multiple first reference points that form the separation section, a second reference point is associated with which the ratio of the distance from the starting position to the first reference point to the distance of the separation section on the first reference line matches the ratio of the distance from the starting position to the second reference point to the distance of the separation section on the second reference line. generating a first projection image in which the first three-dimensional point cloud, the positions of which have been corrected by the second correction unit, is projected onto a projection plane including the reference line; and selecting, from the first projection image, a first line pattern that extends in a direction perpendicular to the reference line and has a length equal to or greater than a predetermined second reference distance;
detecting a second line pattern having a shape identical to that of the first line pattern from a second projection image in which the second three-dimensional point cloud, the position of which has been corrected by the second correction unit, is projected onto the projection surface;
9. The position information processing device according to claim 1, further comprising a third correction unit that calculates a difference in position between the second straight line pattern and the first straight line pattern, and corrects, using the difference in position, the position of a point on the reference straight line that corresponds to the position of the first straight line pattern, and the position of a point included in the first three-dimensional point cloud that is associated with the first reference point that corresponds to the point on the reference straight line. generating a first cross-sectional image representing the first three-dimensional point cloud, the positions of which have been corrected by the third correction unit, the first cross-sectional image including points on the reference line corrected by the third correction unit and included in a cross-section orthogonal to the reference line;
selecting a third line pattern from the first cross-sectional image, the third line pattern extending in a direction perpendicular to the reference line and having a length equal to or greater than a predetermined third reference distance;
generating a second cross-sectional image representing the second three-dimensional point cloud corrected by the second correction unit and included in the cross-section;
10. The position information processing device according to claim 9, further comprising a fourth correction unit that projects the third line pattern onto the second cross-sectional image, rotates the third line pattern around the reference line as a rotation axis, calculates a rotation angle at which a fourth line pattern in the second cross-sectional image whose shape matches that of the third line pattern is detected, and uses the rotation angle to correct angles around the reference line of points included in the first three-dimensional point cloud that are associated with the first reference point corresponding to a point on the reference line included in the first cross-sectional image. The position information processing device according to claim 10, further comprising a fifth correction unit that corrects the positions of points included in the first three-dimensional point cloud associated with the first reference point, the positions of which have been corrected by the fourth correction unit, using the difference in position between the first reference point and a point on the reference line corresponding to the first reference point. Determine a start position where a distance between a first reference line representing a predetermined linear environment set for a first three-dimensional point cloud representing the position of the environment around a first travel path of the first vehicle, the first reference line being set for a first three-dimensional point cloud representing the position of the environment around a first travel path of the first vehicle, the first reference line being set for a second three-dimensional point cloud representing the position of the environment around a second travel path including at least a portion of the same as the first travel path, the second reference line being set for a second three-dimensional point cloud being set for a second vehicle, the second reference line being set for a second three-dimensional point cloud representing the position of the environment around a second travel path including at least a portion of the same as the first travel path, the second reference line being set for a start position where a distance between the first reference line and the second reference line representing the predetermined linear environment set for a second three-dimensional point cloud representing the position of the environment around a second travel path including at least a portion of the same as the first travel path, the second reference line being ... start position where a distance between the first reference line and the second reference line representing the predetermined linear environment set for a second three-dimensional point cloud starts to be greater than a predetermined first reference distance, and a separation section where a distance between the first reference line and the second reference line representing the predetermined linear environment set for a second three-dimensional point cloud is greater than the first reference distance;
associate each of a plurality of first reference points forming the separated section of the first reference line with a point included in the first three-dimensional point cloud;
Corresponding each of the first reference points forming the spaced section of the first reference line to a second reference point forming the spaced section of the second reference line;
for each of the plurality of first reference points forming the separation section of the first reference line, a difference in position between the first reference point and a second reference point associated with the first reference point is calculated, and the difference in position is used to correct the position of a point included in the first three-dimensional point cloud associated with the first reference point;
a reference straight line is generated by arranging each of the plurality of first reference points forming the separated section of the first reference line so as to extend in a predetermined direction from the start position as a starting point, so that the distance between adjacent first reference points is maintained; and a reference straight line is generated by arranging each of the plurality of second reference points forming the separated section of the second reference line so as to maintain the distance between adjacent second reference points, so that the distance between adjacent second reference points is maintained;
for each of a plurality of first reference points forming the separated section of the first reference line, a difference in position between the first reference point and a point on the corresponding reference line is calculated, and the position of a point included in the first three-dimensional point cloud associated with the first reference point corrected by the correction is corrected using the difference in position; and for each of a plurality of second reference points forming the separated section of the second reference line, a difference in position between the second reference point and a point on the corresponding reference line is calculated, and the position of a point included in the second three-dimensional point cloud associated with the second reference point is corrected using the difference in position.
2. A position information processing method executed by a position information processing device.