Coordinate alignment method and apparatus

Abstract

To provide a coordinate alignment method and apparatus capable of aligning a coordinate origin and a coordinate axis in a small order while reducing the number of steps required to align the coordinate system of a point cloud map.SOLUTION: The coordinate alignment method includes: a step of acquiring a point cloud map; a step of extracting a point cloud of a first landmark for determining a coordinate origin in the point cloud map; a step of extracting a point cloud of a second landmark for determining coordinate axes in the point cloud map; an origin alignment step of obtaining a coordinate value in a global coordinate system of a point corresponding to the origin of the global coordinate system in the point cloud map based on the point cloud of the first landmark, and offsetting the point cloud of the point cloud map so that the coordinate value becomes the coordinate origin; and an axis alignment step of obtaining the coordinate axis corresponding to the axis of the global coordinate system in the point cloud map based on the point cloud of the second landmark, and rotating the point cloud map to make the coordinate axis parallel to the axis of the global coordinate system.SELECTED DRAWING: Figure 3

Description

Technical Field

[0001] The present invention relates to a coordinate alignment method and apparatus.

Background Art

[0002] For example, Patent Document 1 describes a technique for processing map data used for self-position estimation of a moving body equipped with an external sensor such as a laser range finder. In self-position estimation of a moving body, a SLAM (simultaneous localization and mapping) method that performs self-position estimation and map creation simultaneously is used. The map is a point cloud map composed of a point cloud indicating the positions of the reflection points of laser beams.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] By the way, in self-position estimation of a moving body using the SLAM method, since the arithmetic processing of self-position estimation is performed in the global coordinate system, if the coordinate system of the point cloud map is deviated from the global coordinate system, an error will occur in the self-position estimation result. Therefore, it is necessary to align the coordinate system of the point cloud map with the global coordinate system. However, when manually aligning the coordinate system of the point cloud map, it is extremely difficult to align the coordinate origin and the coordinate axes on a small order, and there is no reproducibility. For this reason, the self-position estimation result changes depending on the person who aligns the coordinate system of the point cloud map.

[0005] An object of the present invention is to provide a coordinate alignment method and apparatus capable of aligning the coordinate origin and the coordinate axes on a small order while reducing the man-hours for aligning the coordinate system of the point cloud map.

Means for Solving the Problems

[0006] One aspect of the present invention is a coordinate alignment method for aligning the coordinate system of a point cloud map used to estimate the self-position of a moving object using a laser sensor to a global coordinate system, comprising: an acquisition step of acquiring a point cloud map; a first point cloud extraction step of extracting a point cloud of a first landmark for determining the coordinate origin in the point cloud map; a second point cloud extraction step of extracting a point cloud of a second landmark for determining the coordinate axes in the point cloud map; an origin alignment step of aligning the coordinate origin of the point cloud map to the origin of the global coordinate system by determining the coordinate values ​​in the global coordinate system of a point in the point cloud map corresponding to the origin of the global coordinate system based on the point cloud of the first landmark, and offsetting the point cloud of the point cloud map so that the coordinate values ​​become the coordinate origin; and an axis alignment step of aligning the coordinate axes of the point cloud map to the axes of the global coordinate system by determining the coordinate axes in the point cloud map corresponding to the axes of the global coordinate system in the point cloud map based on the point cloud of the second landmark, and rotating the point cloud map so that the coordinate axes are parallel to the axes of the global coordinate system.

[0007] In this coordinate alignment method, based on the point cloud of a first landmark used to determine the coordinate origin in the point cloud map, the coordinate values ​​in the global coordinate system of the point corresponding to the origin of the global coordinate system in the point cloud map are determined. By offsetting the point cloud of the point cloud map so that the coordinate value becomes the coordinate origin, the coordinate origin of the point cloud map aligns with the origin of the global coordinate system. Furthermore, based on the point cloud of a second landmark used to determine the coordinate axes in the point cloud map, the coordinate axes corresponding to the axes of the global coordinate system in the point cloud map are determined. By rotating the point cloud map so that the coordinate axes are parallel to the axes of the global coordinate system, the coordinate axes of the point cloud map align with the axes of the global coordinate system. Here, the process of aligning the coordinate origin of the point cloud map with the origin of the global coordinate system (origin alignment process) and the process of aligning the coordinate axes of the point cloud map with the axes of the global coordinate system (axis alignment process) can be performed automatically. This reduces the effort required to align the coordinate system of the point cloud map, while aligning the coordinate origin and coordinate axes to a small order of magnitude.

[0008] The first landmark is a polygonal or circular feature in the point cloud map, and the second landmark may be multiple features arranged along a specified direction in the point cloud map or a feature extending in a specified direction. In this case, since a feature with a simple shape is used as the first landmark, the coordinate values ​​in the global coordinate system of the point corresponding to the origin of the global coordinate system can be easily determined. Furthermore, by performing line detection on the point cloud of the second landmark, the coordinate axes of the point cloud map can be easily determined.

[0009] The second landmark includes the first landmark, and the axis alignment process may be performed after the origin alignment process. In this case, by obtaining coordinate axes that include the coordinate origin in the point cloud map, the process of aligning the coordinate axes of the point cloud map with the axes of the global coordinate system can be simplified.

[0010] Another aspect of the present invention is a coordinate alignment device for aligning the coordinate system of a point cloud map used to estimate the self-position of a moving object using a laser sensor to a global coordinate system, comprising: a storage unit that stores data of a point cloud map, data of a first landmark point cloud for determining the coordinate origin in the point cloud map, and data of a second landmark point cloud for determining the coordinate axes in the point cloud map; an origin alignment processing unit that aligns the coordinate origin of the point cloud map to the origin of the global coordinate system by determining the coordinate values ​​in the global coordinate system of a point in the point cloud map corresponding to the origin of the global coordinate system based on the point cloud of the first landmark, and offsetting the point cloud of the point cloud map so that the coordinate values ​​become the coordinate origin; an axis alignment processing unit that aligns the coordinate axes of the point cloud map to the axes of the global coordinate system by determining the coordinate axes in the point cloud map corresponding to the axes of the global coordinate system in the point cloud map based on the point cloud of the second landmark, and rotating the point cloud map so that the coordinate axes are parallel to the axes of the global coordinate system; and a storage processing unit that stores the new point cloud map obtained by the origin alignment processing unit and the axis alignment processing unit in the storage unit.

[0011] In such a coordinate alignment device, based on the point cloud of a first landmark used to determine the coordinate origin in the point cloud map, the coordinate values ​​in the global coordinate system of the point corresponding to the origin of the global coordinate system in the point cloud map are determined. By offsetting the point cloud of the point cloud map so that the coordinate value becomes the coordinate origin, the coordinate origin of the point cloud map aligns with the origin of the global coordinate system. Furthermore, based on the point cloud of a second landmark used to determine the coordinate axes in the point cloud map, the coordinate axes corresponding to the axes of the global coordinate system in the point cloud map are determined. By rotating the point cloud map so that the coordinate axes are parallel to the axes of the global coordinate system, the coordinate axes of the point cloud map align with the axes of the global coordinate system. Here, the processes of aligning the coordinate origin of the point cloud map with the origin of the global coordinate system and aligning the coordinate axes of the point cloud map with the axes of the global coordinate system can be performed automatically. This reduces the effort required to align the coordinate system of the point cloud map, while aligning the coordinate origin and coordinate axes to a small order of magnitude. [Effects of the Invention]

[0012] According to the present invention, the effort required to align the coordinate systems of point cloud maps is reduced, while the coordinate origin and coordinate axes can be aligned to a small order of magnitude. [Brief explanation of the drawing]

[0013] [Figure 1] This is a block diagram showing the configuration of a travel control device mounted on a mobile body to which a coordinate alignment method and apparatus according to one embodiment of the present invention are applied. [Figure 2] This diagram schematically shows an example of the environment and point cloud map in which a moving object operates. [Figure 3] This is a flowchart showing the steps of a coordinate alignment method according to one embodiment of the present invention. [Figure 4] Figure 2(b) schematically shows an example of the first and second landmarks in the point cloud data. [Figure 5] This is a block diagram showing the configuration of a coordinate alignment device according to one embodiment of the present invention. [Figure 6] It is a flowchart showing the arithmetic processing procedure executed by the arithmetic processing unit shown in FIG. 5. [Figure 7] It is a diagram showing how the point cloud map shown in FIG. 4 is rotated. [Figure 8] It is a diagram schematically showing a modification example of the first landmark and the second landmark in the point cloud data shown in FIG. 4.

Embodiments for Carrying Out the Invention

[0014] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

[0015] FIG. 1 is a block diagram showing the configuration of a travel control device mounted on a moving body to which the coordinate alignment method and apparatus according to an embodiment of the present invention are applied. In FIG. 1, the travel control device 1 is a device that controls the moving body 2 (see FIG. 2(a)) to travel automatically. The moving body 2 is an industrial vehicle such as a forklift or an automated guided vehicle.

[0016] The travel control device 1 includes a laser sensor 3, a map generation processing unit 4, a point cloud map storage unit 5, a drive unit 6, and a travel control processing unit 7.

[0017] <![CDATA[The laser sensor 3 irradiates laser light toward the surroundings of the moving body 2 and receives the reflected light of the laser light, thereby detecting the distance to an object existing in the surroundings of the moving body 2 and acquiring point cloud detection data. The laser sensor 3 irradiates three-dimensional laser light within a predetermined angular range (for example, 270 degrees) centered on the front of the moving body 2. The point cloud is a collection of reflection points (laser points) of the laser light. As the laser sensor 3, for example, LIDAR (light detection and ranging) or a laser rangefinder is used.]]

[0018] The map generation processing unit 4 is composed of a CPU, a RAM, a ROM, an input / output interface, etc. The map generation processing unit 4 generates a point cloud map of the area where the mobile body 2 travels based on the point cloud detection data of the laser sensor 3, and stores the point cloud map in the point cloud map storage unit 5. The map generation processing unit 4 generates the point cloud map while driving the mobile body 2 manually.

[0019] Fig. 2(a) shows an example of the environment where the mobile body 2 actually travels. Fig. 2(b) shows the point cloud map M of the environment shown in Fig. 2(a). The point cloud map M is a map in which the stationary object A is represented by point clouds. The stationary object A is a pillar, a shelf, a wall, etc. In Fig. 2(b), the point cloud map M is represented in two-dimensional coordinates (XY coordinates) for convenience, but the actual point cloud map M is in three-dimensional coordinates (XYZ coordinates). Also, in Fig. 2, only a rectangular prism is represented as the stationary object A for convenience.

[0020] The drive unit 6 has, for example, although not shown in the figure, a travel motor for driving the mobile body 2 and a steering motor for steering the mobile body 2.

[0021] The travel control processing unit 7 is composed of a CPU, a RAM, a ROM, an input / output interface, etc. The travel control processing unit 7 has a self-position estimation unit 8 and a guidance control unit 9.

[0022] The self-position estimation unit 8 estimates the self-position of the mobile body 2 based on the point cloud detection data of the laser sensor 3 and the point cloud map M stored in the point cloud map storage unit 5. Specifically, the self-position estimation unit 8 estimates the self-position of the mobile body 2 by matching the point cloud detection data and the point cloud map M using, for example, the SLAM (simultaneous localization and mapping) method. SLAM is a self-position estimation technology that performs self-position estimation using sensor data and map data.

[0023] The guidance control unit 9 controls the drive unit 6 to guide the mobile body 2 to the target position based on the self-position of the mobile body 2 estimated by the self-position estimation unit 8.

[0024] Incidentally, the self-position estimation calculation by the self-position estimation unit 8 is performed in the global coordinate system. The global coordinate system is the real-world coordinate system. Therefore, as shown in Figure 2, if the coordinate system of the point cloud map M is misaligned with the global coordinate system, a difference will occur between the estimated position of the moving object 2 obtained by the self-position estimation unit 8 and the actual position of the moving object 2. Note that G0 in Figure 2 is the origin of the global coordinate system, and Jx0 in Figure 2 is the X-axis of the global coordinate system. G in Figure 2 is the origin of the coordinate system of the point cloud map M, and Jx in Figure 2 is the X-axis of the coordinate system of the point cloud map M.

[0025] For example, even if the moving object 2 actually moves from coordinate position (2,2,0) to coordinate position (4,2,0) in the X-axis direction, the self-position estimation result obtained by the self-position estimation unit 8 will incorrectly indicate that the moving object 2 moved from coordinate position (1,1,0) to coordinate position (2.5,1.5,0).

[0026] Therefore, it is necessary to align the coordinate system of the point cloud map M with the global coordinate system. In this embodiment, as shown in Figure 2, the lower left corner of the rectangular prism located in the lower left of the page is set as the origin, and the direction of the arrangement of the rectangular prisms along the left and right sides of the page is set as the X-axis, thereby aligning the coordinate system of the point cloud map M with the global coordinate system.

[0027] Figure 3 is a flowchart showing the steps of a coordinate alignment method according to one embodiment of the present invention. The coordinate alignment method of this embodiment is a method for aligning the coordinate system of a point cloud map M with the global coordinate system.

[0028] This coordinate alignment of the point cloud map M is performed only once, before the mobile body 2 is automatically driven by the driving control device 1. At this time, the operator uses the personal computer 11 (see Figure 5) to perform the coordinate alignment of the point cloud map M.

[0029] In Figure 3, first, an acquisition process is performed to acquire the point cloud map M stored in the point cloud map storage unit 5 of the driving control device 1 (process S101). Specifically, the point cloud map M is imported into the personal computer 11 by the operator's manual operation, and the point cloud map M is displayed on the monitor screen of the personal computer 11.

[0030] Next, a process is performed to extract the point cloud of the first landmark L1 for determining the coordinate origin in the point cloud map M (step S102). The first landmark L1 is a polygonal feature in the point cloud map M. Here, as shown in Figure 4, the first landmark L1 is a quadrilateral feature (corresponding to a quadrilateral prism) located in the lower left of the point cloud map M.

[0031] Specifically, the operator extracts the point cloud of the first landmark L1 by manually operating point cloud processing GUI (Graphical User Interface) software such as CloudCompare while viewing the monitor screen of the PC 11. The extracted point cloud data of the first landmark L1 is stored in the landmark storage unit 13 of the PC 11, which will be described later.

[0032] Next, a process is performed to extract the point cloud of the second landmark L2 in order to determine the coordinate axes (X axis, Y axis, Z axis) in the point cloud map M (step S103). The second landmark L2 consists of multiple features arranged along a specified direction in the point cloud map M. Here, the second landmark L2 consists of two quadrilateral features (corresponding to a rectangular prism) located in the lower left and lower right of the point cloud map M, as shown in Figure 4. The second landmark L2 includes the first landmark L1.

[0033] Specifically, the operator extracts the point cloud of the second landmark L2 by manually operating the point cloud processing GUI software while viewing the monitor screen of the PC 11, similar to the extraction of the point cloud of the first landmark L1. The extracted point cloud data of the second landmark L2 is stored in the landmark storage unit 13 (described later) of the PC 11.

[0034] Next, the point cloud map M is coordinate-aligned using the point cloud of the first landmark L1 extracted in step S102 and the point cloud of the second landmark L2 extracted in step S103 (step S104). The coordinate-alignment of the point cloud map M is performed automatically by the coordinate-alignment device 10 described below.

[0035] Figure 5 is a block diagram showing the configuration of a coordinate alignment device according to one embodiment of the present invention. In Figure 5, the coordinate alignment device 10 of this embodiment is a device that implements the coordinate alignment method described above. In other words, the coordinate alignment device 10 is a device that aligns the coordinate system of the point cloud map M with the global coordinate system, and is composed of a personal computer 11 (PC).

[0036] The personal computer 11 includes a point cloud map storage unit 12, a landmark storage unit 13, and a processing unit 14.

[0037] The point cloud map storage unit 12 stores the data of the point cloud map M before the coordinate alignment process is performed. In other words, the point cloud map storage unit 12 stores the data of the point cloud map M generated by the map generation processing unit 4 of the driving control device 1 described above.

[0038] As described above, the landmark memory unit 13 stores the point cloud data of the first landmark L1 and the point cloud data of the second landmark L2.

[0039] The arithmetic processing unit 14 consists of a CPU, RAM, ROM, and an input / output interface, etc. The arithmetic processing unit 14 has an origin alignment processing unit 15, an axis alignment processing unit 16, and a storage processing unit 17.

[0040] The origin alignment processing unit 15 determines the coordinate values ​​in the global coordinate system of the point corresponding to the origin of the global coordinate system in the point cloud map M based on the point cloud of the first landmark L1 stored in the landmark memory unit 13, and offsets the point cloud of the point cloud map M so that the coordinate values ​​become the coordinate origin, thereby aligning the coordinate origin of the point cloud map M with the origin of the global coordinate system.

[0041] The axis alignment processing unit 16 determines the coordinate axes in the point cloud map M that correspond to the axes of the global coordinate system based on the point cloud of the second landmark L2 stored in the landmark memory unit 13, and rotates the point cloud map M so that these coordinate axes are parallel to the axes of the global coordinate system, thereby aligning the coordinate axes of the point cloud map M with the axes of the global coordinate system.

[0042] The storage processing unit 17 stores the new point cloud map obtained by the origin alignment processing unit 15 and the axis alignment processing unit 16 in the point cloud map storage unit 12.

[0043] Figure 6 is a flowchart showing the calculation procedure performed by the calculation unit 14. In Figure 6, the calculation unit 14 acquires the data of the point cloud map M stored in the point cloud map storage unit 12 (procedure S111). The calculation unit 14 also acquires the point cloud data of the first landmark L1 and the point cloud data of the second landmark L2 stored in the landmark storage unit 13 (procedure S112).

[0044] Next, the arithmetic processing unit 14 processes the point cloud of the first landmark L1 to calculate the coordinate values ​​in the global coordinate system of the point to be used as the coordinate origin in the point cloud map M (procedure S113). The point to be used as the coordinate origin in the point cloud map M is the point corresponding to the origin of the global coordinate system in the point cloud map M. Here, as mentioned above, the origin of the global coordinate system is the lower left corner of the rectangular prism located in the lower left (see G0 in Figure 2(a)). Therefore, the point corresponding to the origin of the global coordinate system in the point cloud map M is point G in Figure 7. Alternatively, the point corresponding to the origin of the global coordinate system in the point cloud map M may be a point located near point G in Figure 7. Methods for obtaining the coordinate values ​​of the origin include the Hough transform.

[0045] Next, the arithmetic processing unit 14 offsets all points in the point cloud map M so that the coordinate value of the point to be designated as the coordinate origin becomes the coordinate origin (procedure S114). As a result, the coordinate origin of the point cloud map M will approximately coincide with the origin of the global coordinate system. For example, if the coordinate value of the point to be designated as the coordinate origin is (1,2,0), then all points in the point cloud map M will be offset by (-1,-2,0) so that the coordinate value of that point becomes (0,0,0).

[0046] Next, the arithmetic processing unit 14 calculates the coordinate axes of the point cloud map M corresponding to the axes of the global coordinate system by performing geometrical line detection on the point cloud of the second landmark L2 and calculating the equation of the line (procedure S115). The coordinate axes of the point cloud map M corresponding to the axes of the global coordinate system are coordinate axes that coincide with the axes of the global coordinate system or coordinate axes that are close to the axes of the global coordinate system. At this time, the arithmetic processing unit 14 calculates coordinate axes that pass through the coordinate origin of the point cloud map M, as shown by Jx in Figure 7.

[0047] Methods for calculating the equation of a straight line include RANSAC (Random Sample Consensus) and the least squares method. For example, after removing outliers from the point cloud of the second landmark L2 using RANSAC to obtain the equation of a straight line, the least squares method can be used to obtain the equation of a straight line that eliminates the variation between trials. This yields a straight line that resembles a coordinate axis.

[0048] Next, the arithmetic processing unit 14 rotates the point cloud map M so that its coordinate axes are parallel to the axes of the global coordinate system (procedure S116). As a result, the coordinate axes of the point cloud map M almost coincide with the axes of the global coordinate system.

[0049] Specifically, as shown in Figure 7, if the equation of the line to be used as the X-axis is y = ax + b, the angle θ such that y = 0 is found, and the entire point cloud map is rotated by the angle θ. The angle θ is the angle between the X-axis (Jx) of the point cloud map M and the X-axis (Jx0) of the global coordinate system. In this case, since the slope of the equation of the line is a, θ = atan(a). Therefore, a rotation matrix is ​​used to rotate the entire point cloud map M by the angle θ. This calculation process is performed for all three axes (X, Y, Z).

[0050] As a result, a new point cloud map is obtained in which the coordinate origin and coordinate axes are aligned with the global coordinate system.

[0051] Next, the arithmetic processing unit 14 overwrites the point cloud map in the point cloud map storage unit 12 with the new point cloud map (procedure S117). As a result, when the mobile body 2 is driven automatically, its own position is estimated using the new point cloud map.

[0052] In the above, the origin alignment processing unit 15 executes the above steps S111 to S114. The axis alignment processing unit 16 executes the above steps S111, S112, S115, and S116. The storage processing unit 17 executes the above step S117.

[0053] Furthermore, step S101 in Figure 3 is an acquisition step for acquiring the point cloud map M. Step S102 in Figure 3 is a first point cloud extraction step for extracting the point cloud of the first landmark L1 for determining the coordinate origin in the point cloud map M. Step S103 in Figure 3 is a second point cloud extraction step for extracting the point cloud of the second landmark L2 for determining the coordinate axes in the point cloud map M. Step S104 in Figure 3 includes an origin alignment step performed by the origin alignment processing unit 15 and an axis alignment step performed by the axis alignment processing unit 16.

[0054] As described above, in this embodiment, based on the point cloud of the first landmark L1 for determining the coordinate origin in the point cloud map M, the coordinate values ​​in the global coordinate system of the point corresponding to the origin of the global coordinate system in the point cloud map M are determined, and the point cloud of the point cloud map M is offset so that the coordinate value becomes the coordinate origin, thereby aligning the coordinate origin of the point cloud map M with the origin of the global coordinate system. Furthermore, based on the point cloud of the second landmark L2 for determining the coordinate axes in the point cloud map M, the coordinate axes corresponding to the axes of the global coordinate system in the point cloud map M are determined, and the point cloud map M is rotated so that the coordinate axes are parallel to the axes of the global coordinate system, thereby aligning the coordinate axes of the point cloud map M with the axes of the global coordinate system. Here, the process of aligning the coordinate origin of the point cloud map M with the origin of the global coordinate system and the process of aligning the coordinate axes of the point cloud map M with the axes of the global coordinate system can be performed automatically. This reduces the man-hours required to align the coordinate system of the point cloud map M, while aligning the coordinate origin and coordinate axes on a small order of magnitude. As a result, it becomes possible to reduce the self-position estimation error of the moving object 2 due to the shift in the coordinate origin and coordinate axes of the point cloud map M relative to the global coordinate system.

[0055] Furthermore, in this embodiment, the first landmark L1 is a polygonal feature in the point cloud map M, and the second landmark L2 is a plurality of features arranged along a specified direction in the point cloud map M. In this case, since a feature with a simple shape is used as the first landmark L1, the coordinate values ​​in the global coordinate system of the point corresponding to the origin of the global coordinate system can be easily obtained. In addition, by performing line detection on the point cloud of the second landmark L2, the coordinate axes of the point cloud map M can be easily obtained.

[0056] Furthermore, in this embodiment, the second landmark L2 includes the first landmark L1, and the axis alignment process is performed after the origin alignment process. In this case, by obtaining coordinate axes that include the coordinate origin in the point cloud map M, the process of aligning the coordinate axes of the point cloud map M with the axes of the global coordinate system can be simplified.

[0057] It should be noted that the present invention is not limited to the embodiments described above. For example, in the above embodiment, the lower left corner of the first landmark L1 (a quadrilateral feature) located in the lower left of the point cloud map M is set as the coordinate origin, but the invention is not limited to such a form. For example, if the origin of the global coordinate system is another corner of the quadrilateral prism, then that corner of the first landmark L1 in the point cloud map M is set as the coordinate origin. Also, if the origin of the global coordinate system is the horizontal center of the quadrilateral prism, then the center of the first landmark L1 in the point cloud map M is set as the coordinate origin.

[0058] Furthermore, in the above embodiment, the first landmark L1 is a polygonal feature in the point cloud map M, but it is not limited to such a form. The first landmark L1 may be a circular feature corresponding to a cylinder in the point cloud map M, for example, as shown in Figure 8(a). In this case, the center of the first landmark L1 in the point cloud map M is set as the coordinate origin. In this case as well, a feature with a simple shape is used as the first landmark L1. Therefore, the coordinate values ​​in the global coordinate system of the point corresponding to the origin of the global coordinate system can be easily obtained.

[0059] Furthermore, in the above embodiment, the second landmark L2 is a plurality of features arranged along a specified direction in the point cloud map M, but it is not limited to such a form. The second landmark L2 may be a feature extending in a specified direction in the point cloud map M, for example, as shown in Figure 8(b). Such a feature corresponds to a wall, shelf, etc. In this case as well, a line resembling a coordinate axis will be obtained by line detection. Therefore, the coordinate axes of the point cloud map M can be easily determined.

[0060] Furthermore, in the above embodiment, the axis alignment process is performed after the origin alignment process, but the embodiment is not limited to this configuration, and the origin alignment process may be performed after the axis alignment process. In this case, the second landmark L2 may include the first landmark L1, or the first landmark L1 and the second landmark L2 may be completely different.

[0061] Furthermore, in the above embodiment, a point cloud map storage unit 12 is provided to store the data of the point cloud map M before the coordinate alignment process is performed, and a landmark storage unit 13 is provided to store the point cloud data of the first landmark L1 and the second landmark L2, but the system is not limited to such an configuration. The data of the point cloud map M and the point cloud data of the first landmark L1 and the second landmark L2 may be stored in the same storage unit.

[0062] Furthermore, although the above embodiment involves coordinate alignment of the 3D point cloud map M, the system is not limited to this configuration. When a laser sensor 3 that emits 2D laser light is used, coordinate alignment of the 2D point cloud map M is performed.

[0063] Furthermore, in the above embodiment, the coordinate alignment of the point cloud map M is performed when a new point cloud map M is created based on the point cloud detection data of the laser sensor 3, but the embodiment is not limited to this. For example, the coordinate alignment of the point cloud map M may also be performed when a part of the point cloud map M is modified based on the point cloud detection data of the laser sensor 3.

[0064] Furthermore, in the above embodiment, the first point cloud extraction step, which extracts the point cloud of the first landmark L1 in the point cloud map M, and the second point cloud extraction step, which extracts the point cloud of the second landmark L2 in the point cloud map M, are performed using external software through manual operation by an operator, but the system is not limited to this configuration. The first point cloud extraction step and the second point cloud extraction step may be performed automatically by the internal software of the personal computer 11.

[0065] Furthermore, in the above embodiment, the point cloud map M is acquired by the operator manually importing it into the personal computer 11, but the system is not limited to this configuration. For example, the point cloud map M may be acquired by automatically sending it from the point cloud map storage unit 5 of the driving control device 1 to the personal computer 11 via wireless communication. [Explanation of Symbols]

[0066] 2...Moving body, 3...Laser sensor, 10...Coordinate alignment device, 12...Point cloud map storage unit (storage unit), 13...Landmark storage unit (storage unit), 15...Origin alignment processing unit, 16...Axis alignment processing unit, 17...Storage processing unit, L1...First landmark, L2...Second landmark, M...Point cloud map.

Claims

1. A coordinate alignment method performed by a computer and used to estimate the self-position of a moving object using a laser sensor, which aligns the coordinate system of a point cloud map in which stationary objects are represented as point clouds with a global coordinate system, A first point cloud extraction step is performed to extract a point cloud of a first landmark for determining the coordinate origin in the point cloud map, A second point cloud extraction step is performed to extract a point cloud of a second landmark for determining the coordinate axes in the aforementioned point cloud map. An origin alignment step is performed to align the coordinate origin of the point cloud map with the origin of the global coordinate system by determining the coordinate value of the point to be the coordinate origin in the point cloud map based on the point cloud of the first landmark, and offsetting the point cloud of the point cloud map so that the coordinate value becomes the coordinate origin, The process includes determining coordinate axes in the point cloud map that correspond to the axes of the global coordinate system based on the point cloud of the second landmark, and rotating the point cloud map so that the coordinate axes are parallel to the axes of the global coordinate system, thereby aligning the coordinate axes of the point cloud map with the axes of the global coordinate system. The stationary object has multiple columns, The first landmark is a polygonal or circular feature corresponding to the column in the point cloud map, The second landmark is a coordinate alignment method in which a plurality of polygonal or circular features are arranged along a specified direction and correspond to the columns in the point cloud map.

2. The second landmark includes the first landmark, The coordinate alignment method according to claim 1, wherein the axis alignment step is performed after the origin alignment step is performed.

3. A coordinate alignment device used when estimating the self-position of a moving object using a laser sensor, which aligns the coordinate system of a point cloud map in which stationary objects are represented as point clouds with a global coordinate system, A storage unit that stores the data of the point cloud map, the point cloud data of a first landmark for determining the coordinate origin in the point cloud map, and the point cloud data of a second landmark for determining the coordinate axes in the point cloud map. An origin alignment processing unit that, based on the point cloud of the first landmark, determines the coordinate value in the global coordinate system of the point to be the coordinate origin in the point cloud map, and offsets the point cloud of the point cloud map so that the coordinate value becomes the coordinate origin, thereby aligning the coordinate origin of the point cloud map with the origin of the global coordinate system, An axis alignment processing unit that determines the coordinate axes in the point cloud map that correspond to the axes of the global coordinate system based on the point cloud of the second landmark, and rotates the point cloud map so that the coordinate axes are parallel to the axes of the global coordinate system, thereby aligning the coordinate axes of the point cloud map with the axes of the global coordinate system, The system includes a storage processing unit that stores a new point cloud map obtained by the origin alignment processing unit and the axis alignment processing unit in the storage unit. The stationary object has multiple columns, The first landmark is a polygonal or circular feature corresponding to the column in the point cloud map, The second landmark is a coordinate alignment device which corresponds to the column in the point cloud map and is a plurality of polygonal or circular features arranged along a specified direction.

Images