Information processing device, control method, program and storage medium

Abstract

To provide an information processing device capable of suitably performing downsampling.SOLUTION: A controller 13 of an on-vehicle device 1 acquires point group data output by a lidar 2. Then, the controller 13 generates processing point group data by downsampling the point group data. Then, the controller 13 matches the processing point group data with voxel data VD representing the position of an object for each voxel which is a unit area, thereby associating measurement points and each voxel constituting the processing point group data. Then, the controller 13 changes the size of the downsampling to be performed next based on the number of corresponding measurement points Nc, which is the number of measurement points associated with voxels among the measurement points of the processing point group data.SELECTED DRAWING: Figure 12

Description

[Technical Field]

[0001] The present invention relates to downsampling of data used for position estimation. [Background technology]

[0002] Conventionally, there has been known a technique for estimating a vehicle's own position by comparing (matching) shape data of surrounding objects measured using a measurement device such as a laser scanner with map information in which the shapes of surrounding objects are stored in advance. For example, Patent Document 1 discloses an autonomous mobile system that determines whether a detected object in a voxel obtained by dividing a space according to a predetermined rule is a stationary object or a moving object, and matches the map information with the measurement data for voxels in which a stationary object exists. Furthermore, Patent Document 2 discloses a scan matching method that estimates a vehicle's own position by comparing voxel data including the mean vector and covariance matrix of stationary objects for each voxel with point cloud data output by a lidar. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] International Publication WO2013 / 076829 [Patent Document 2] International Publication No. WO2018 / 221453 Summary of the Invention [Problem to be solved by the invention]

[0004] As in Patent Document 2, when estimating vehicle position by comparing voxel data with point cloud data output by a LIDAR, a downsampling process is performed, which is a process of thinning out the LIDAR point cloud data by averaging it over a space divided into predetermined sizes. This downsampling process makes the density of the point cloud data uniform, with areas close to the vehicle being dense and areas farther away being sparse, and reduces the overall number of point cloud data, thereby improving the accuracy of self-position estimation and reducing calculation time. However, in spaces with few objects around the vehicle, the number of point cloud data detected by the LIDAR is small, so the downsampling process can significantly reduce the number of point cloud data used for calculation, resulting in a problem of reduced accuracy and robustness of self-position estimation.

[0005] The present invention has been made to solve the above-mentioned problems, and a main object of the present invention is to provide an information processing device that can preferably perform downsampling. [Means for solving the problem]

[0006] The claimed invention is an acquisition means for acquiring point cloud data output by a measurement device; a downsampling processing means for generating processed point cloud data by downsampling the point cloud data; a correlation means for correlating the processing point cloud data with voxel data representing the position of an object for each voxel, which is a unit area, to correlate the measurement points constituting the processing point cloud data with each of the voxels, The downsampling processing means is an information processing device that changes the size of the downsampling to be performed next based on the number of corresponding measurement points, which is the number of measurement points associated with the voxel among the measurement points.

[0007] The claimed invention also includes: A computer-implemented control method comprising: Acquire the point cloud data output by the measuring device, generating processed point cloud data by downsampling the point cloud data; By comparing the processing point cloud data with voxel data representing the position of an object for each voxel, which is a unit area, the measurement points constituting the processing point cloud data are associated with each of the voxels; This is a control method in which the size of the downsampling to be performed next is changed based on the number of corresponding measurement points, which is the number of measurement points among the measurement points that are associated with the voxel.

[0008] The claimed invention also includes: Acquire the point cloud data output by the measuring device, generating processed point cloud data by downsampling the point cloud data; By comparing the processing point cloud data with voxel data representing the position of an object for each voxel, which is a unit area, the measurement points constituting the processing point cloud data are associated with each of the voxels; The program causes a computer to execute a process of changing the size of the downsampling to be executed next, based on the number of corresponding measurement points, which is the number of measurement points among the measurement points that are associated with the voxel. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is a schematic configuration diagram of a driving assistance system. [Figure 2] FIG. 2 is a block diagram showing the functional configuration of the vehicle-mounted device. [Figure 3] 2 shows an example of functional blocks of the downsampling processing unit 14. [Figure 4] FIG. 10 is a diagram showing the first execution results in which downsampling is performed using different downsampling sizes. [Figure 5] FIG. 10 is a diagram showing the second execution results in which downsampling is performed using different downsampling sizes. [Figure 6] FIG. 10 is a diagram showing the third execution result in which downsampling is performed using different downsampling sizes. [Figure 7] An overhead view of the vehicle and its surroundings is shown. [Figure 8] 2 is a diagram showing the vehicle position to be estimated by the vehicle position estimation unit in two-dimensional orthogonal coordinates. FIG. [Figure 9] 1 shows an example of a schematic data structure of voxel data. [Figure 10] 3 is an example of a functional block of a vehicle position estimation unit. [Figure 11] This shows the positional relationship between voxels where voxel data exists and measurement points that indicate positions near these voxels on a two-dimensional plane in the world coordinate system. [Figure 12] 10 is an example of a flowchart showing a procedure of a process relating to vehicle position estimation. [Figure 13] The results of downsampling to a cubic grid size with a side length of 0.5 m and estimating the vehicle position using NDT processing are shown below. [Figure 14] The results of downsampling to a cubic grid size with a side length of 0.5 m and estimating the vehicle position using NDT processing are shown below. [Figure 15] The results of downsampling to a cubic grid size of 2.0 m on a side and estimating the vehicle position using NDT processing are shown below. [Figure 16] The results of downsampling to a cubic grid size of 2.0 m on a side and estimating the vehicle position using NDT processing are shown below. [Figure 17] The results of vehicle position estimation using downsampling and NDT processing according to the processing procedure shown in FIG. 12 are shown. [Figure 18] The results of vehicle position estimation using downsampling and NDT processing according to the processing procedure shown in FIG. 12 are shown. [Figure 19] This shows the results of estimating the vehicle position by downsampling and NDT processing according to the processing procedure shown in FIG. 12 when occlusion by another vehicle occurs. [Figure 20]19, the results are shown for the case where the downsampling size is determined so that the number of measurement points after downsampling falls within the range of 600 to 800. [Figure 21] The results of estimating the vehicle position using NDT processing based on Modification 1 are shown, with the downsampling size in the traveling direction fixed at 0.5 m and the downsampling sizes in the horizontal and vertical directions adaptively changed. [Figure 22] The results of estimating the vehicle position using NDT processing based on Modification 1 are shown, with the downsampling size in the traveling direction fixed at 0.5 m and the downsampling sizes in the horizontal and vertical directions adaptively changed. DETAILED DESCRIPTION OF THE INVENTION

[0010] According to a preferred embodiment of the present invention, an information processing device includes an acquisition means for acquiring point cloud data output by a measurement device, a downsampling processing means for generating processed point cloud data by downsampling the point cloud data, and a correspondence means for matching measurement points constituting the processed point cloud data with each of the voxels by comparing the processed point cloud data with voxel data representing the position of an object for each voxel, which is a unit area, and the downsampling processing means changes the size of the downsampling to be performed next based on the number of corresponding measurement points, which is the number of measurement points of the processed point cloud data associated with the voxel among the measurement points. According to this aspect, the information processing device can adaptively change the downsampling size based on the number of corresponding measurement points, which is the number of measurement points of the processed point cloud data associated with the voxel in which voxel data exists, and suitably optimize the number of corresponding measurement points.

[0011] In one aspect of the information processing device, the downsampling processing means fixes the size in a first direction among directions for dividing space in the downsampling, and changes the size in directions other than the first direction based on the number of corresponding measurement points. In a preferred example, the first direction is the traveling direction of the measuring device. With this aspect, the information processing device can fix the downsampling size in directions where resolution of the processing point cloud data is required, thereby preferably ensuring resolution.

[0012] In another aspect of the information processing device, the downsampling processing means changes the size based on a comparison between the number of corresponding measurement points and a target range of the number of corresponding measurement points. This aspect allows the information processing device to accurately change the downsampling size.

[0013] In another aspect of the information processing device, the downsampling processing means increases the size by a predetermined rate or a predetermined value when the number of corresponding measurement points is greater than the upper limit of the target range, decreases the size by a predetermined rate or a predetermined value when the number of corresponding measurement points is less than the lower limit of the target range, and maintains the size when the number of corresponding measurement points is within the target range. This aspect allows the information processing device to accurately change the downsampling size so that the number of corresponding measurement points is maintained within the target range.

[0014] In another aspect of the information processing device, the downsampling processing means determines a reference size for the next downsampling based on the number of corresponding measurement points, and determines the size for each direction based on the reference size and a ratio to the reference size set for each direction in which space is divided in the downsampling. This aspect allows the information processing device to suitably determine the downsampling size for each direction using a ratio according to the accuracy, importance, etc. of each direction.

[0015] In another aspect of the information processing device, the information processing device further includes a position estimation unit that estimates the position of a moving object equipped with the measurement device based on a comparison result between voxel data of the voxels that have been matched by the matching unit and the measurement points that have been matched to the voxels. In this aspect, the information processing device can preferably achieve both accuracy in position estimation and a short processing time by using processed point cloud data that has been generated by adaptively changing the downsampling size for position estimation.

[0016] In another aspect of the information processing device, the downsampling processing means changes the size based on the processing times of the downsampling processing means and the position estimation means. In a preferred example, the downsampling processing means changes the size based on a comparison between the processing time and a target range for the processing time. This aspect allows the information processing device to suitably set the downsampling size so that the processing times of the downsampling processing means and the position estimation means are appropriate.

[0017] According to another preferred embodiment of the present invention, there is provided a control method executed by an information processing device, which includes acquiring point cloud data output by a measurement device, generating processed point cloud data by downsampling the point cloud data, matching the processed point cloud data with voxel data representing the position of an object for each voxel, which is a unit area, to associate measurement points constituting the processed point cloud data with each of the voxels, and changing the size of the downsampling to be executed next based on the number of corresponding measurement points, which is the number of measurement points among the measurement points associated with the voxel. By executing this control method, the information processing device can adaptively change the downsampling size based on the number of corresponding measurement points, and suitably optimize the number of corresponding measurement points.

[0018] According to yet another preferred embodiment of the present invention, a program causes a computer to acquire point cloud data output by a measurement device, generate processing point cloud data by downsampling the point cloud data, match the processing point cloud data with voxel data representing the position of an object for each voxel, which is a unit area, to associate measurement points constituting the processing point cloud data with each of the voxels, and change the size of the downsampling to be performed next based on the number of corresponding measurement points, which is the number of measurement points associated with the voxel. By executing this program, the computer can adaptively change the downsampling size based on the number of corresponding measurement points and suitably optimize the number of corresponding measurement points. Preferably, the program is stored in a storage medium. [Example]

[0019] Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings. For convenience, in this specification, a character with "^" or "-" added above any symbol will be referred to as "A^" or "A - " (where "A" is any letter).

[0020] (1) Overview of the driving assistance system 1 shows a schematic configuration of a driving assistance system according to this embodiment. The driving assistance system includes an on-board device 1 that moves together with a vehicle, which is a moving body, a Lidar (Light Detection and Ranging or Laser Illuminated Detection and Ranging) 2, a gyro sensor 3, a vehicle speed sensor 4, and a GPS receiver 5.

[0021] The vehicle-mounted device 1 is electrically connected to a lidar 2, a gyro sensor 3, a vehicle speed sensor 4, and a GPS receiver 5, and estimates the position of the vehicle (also referred to as the "vehicle position") on which the vehicle-mounted device 1 is installed based on the outputs of these sensors. Based on the estimated vehicle position, the vehicle-mounted device 1 performs automatic driving control of the vehicle so that the vehicle travels along a route to a set destination. The vehicle-mounted device 1 stores a map database (DB: DataBase) 10 including voxel data "VD." The voxel data VD is data that records position information of stationary structures for each voxel, which represents a cube (regular lattice), the smallest unit of three-dimensional space. The voxel data VD includes data that represents measured point cloud data of stationary structures within each voxel using a normal distribution, and is used for scan matching using the normal distribution transform (NDT), as described below. The vehicle-mounted device 1 estimates at least the vehicle's planar position and yaw angle using NDT scan matching. The vehicle-mounted device 1 may further estimate the vehicle's height position, pitch angle, and roll angle. Unless otherwise specified, the vehicle position is assumed to include the vehicle's attitude angle, such as the yaw angle, to be estimated.

[0022] The LIDAR 2 emits a pulsed laser beam over a predetermined range of angles in the horizontal and vertical directions to discretely measure the distance to an object in the external environment and generate three-dimensional point cloud data indicating the object's location. The LIDAR 2 includes an irradiator that irradiates laser light while changing the irradiation direction, a light-receiving unit that receives the reflected (scattered) light of the irradiated laser beam, and an output unit that outputs scan data (points constituting the point cloud data, hereafter referred to as "measurement points") based on the light-receiving signal output by the light-receiving unit. The measurement points are generated based on the irradiation direction corresponding to the laser beam received by the light-receiving unit and the response delay time of the laser beam determined based on the light-receiving signal. Generally, the closer the distance to the object, the higher the accuracy of the LIDAR's distance measurement value; and the farther the distance, the lower the accuracy. The LIDAR 2, gyro sensor 3, vehicle speed sensor 4, and GPS receiver 5 each provide output data to the vehicle-mounted device 1.

[0023] The on-board device 1 is an example of an "information processing device" in the present invention, and the lidar 2 is an example of a "measurement device" in the present invention. Note that the driving assistance system may have an inertial measurement unit (IMU) instead of or in addition to the gyro sensor 3, which measures the acceleration and angular velocity of the measurement vehicle in three axial directions.

[0024] (2) In-vehicle device configuration 2 is a block diagram showing an example of the hardware configuration of the vehicle-mounted device 1. The vehicle-mounted device 1 mainly includes an interface 11, a memory 12, and a controller 13. These elements are interconnected via a bus line.

[0025] The interface 11 performs interface operations related to the exchange of data between the in-vehicle device 1 and external devices. In this embodiment, the interface 11 acquires output data from sensors such as the lidar 2, the gyro sensor 3, the vehicle speed sensor 4, and the GPS receiver 5, and supplies the data to the controller 13. The interface 11 also supplies signals related to vehicle driving control generated by the controller 13 to an electronic control unit (ECU) of the vehicle. The interface 11 may be a wireless interface such as a network adapter for wireless communication, or may be a hardware interface for connecting to an external device via a cable or the like. The interface 11 may also perform interface operations with various peripheral devices such as an input device, a display device, and a sound output device.

[0026] The memory 12 is configured by various types of volatile and non-volatile memory, such as a RAM (Random Access Memory), a ROM (Read Only Memory), a hard disk drive, and a flash memory. The memory 12 stores programs for the controller 13 to execute predetermined processes. The programs executed by the controller 13 may be stored in a storage medium other than the memory 12.

[0027] The memory 12 also stores target range information 9 and a map DB 10 including voxel data VD.

[0028] The target range information 9 is information used to set the downsampling size when downsampling is performed on the point cloud data obtained when the lidar 2 performs one scanning cycle. Specifically, the target range information 9 indicates the target range of the number of measurement points (also referred to as the "number of corresponding measurement points Nc") that are associated with the voxel data VD in the point cloud data after downsampling in the NDT matching performed at each processing time based on the scanning cycle of the lidar 2. Hereinafter, the target range of the number of corresponding measurement points Nc indicated by the target range information 9 will be referred to as the "target range R Nc " is also called.

[0029] At least one of the target range information 9 and the map DB 10 may be stored in a storage device external to the vehicle-mounted device 1, such as a hard disk connected to the vehicle-mounted device 1 via the interface 11. The storage device may be a server device that communicates with the vehicle-mounted device 1. The storage device may also be composed of multiple devices. The map DB 10 may also be updated periodically. In this case, for example, the controller 13 receives partial map information related to the area to which the vehicle position belongs from a server device that manages map information via the interface 11, and reflects the information in the map DB 10.

[0030] The controller 13 includes one or more processors such as a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), and a TPU (Tensor Processing Unit), and controls the entire in-vehicle device 1. In this case, the controller 13 executes a program stored in the memory 12 or the like to perform processing related to vehicle position estimation.

[0031] Moreover, functionally, the controller 13 has a downsampling processing unit 14 and a vehicle position estimation unit 15. The controller 13 functions as an "acquisition means," a "downsampling processing means," a "correlation means," a computer that executes a program, and the like.

[0032] The downsampling processing unit 14 performs downsampling on the point cloud data output from the lidar 2 to generate point cloud data (also referred to as "processed point cloud data") that has been corrected so as to reduce the number of measurement points. In this downsampling, the downsampling processing unit 14 averages the point cloud data within a space divided into a predetermined size to generate processed point cloud data with a reduced number of measurement points. In this case, the downsampling processing unit 14 calculates the processed point cloud data by calculating the number of corresponding measurement points Nc in NDT matching at each processing time within the target range R Nc The predetermined size for downsampling is adaptively set so that the image falls within the range.

[0033] The vehicle position estimation unit 15 estimates the vehicle position by performing NDT-based scan matching (NDT scan matching) based on the processed point cloud data generated by the downsampling processing unit 14 and the voxel data VD corresponding to the voxels to which the point cloud data belongs.

[0034] (3) Setting the downsampling size Next, a detailed description will be given of the processing of the downsampling processing unit 14. In summary, the downsampling processing unit 14 calculates the number of corresponding measurement points Nc and the target range R indicated by the target range information 9 at each processing time. Nc and after downsampling, the downsampling size at the next processing time is set according to the result of the comparison. Nc The downsampling size is adaptively determined so as to maintain the pixel size within the range.

[0035] Fig. 3 shows an example of the functional blocks of the downsampling processing unit 14. Functionally, the downsampling processing unit 14 has a downsampling block 16, a corresponding measurement point number acquisition block 17, a comparison block 18, and a downsampling size setting block 19. Note that in Fig. 3, blocks where data is exchanged are connected by solid lines, but the combination of blocks where data is exchanged is not limited to that shown in Fig. 3. The same applies to other functional block diagrams described later.

[0036] The downsampling block 16 generates processed point cloud data for each processing time by downsampling the point cloud data for one scanning cycle generated by the LIDAR 2. In this case, the downsampling block 16 divides space into grids based on the set downsampling size and calculates representative points of the measurement points of the LIDAR 2 point cloud data included in each grid. In this case, the space is, for example, a space of a three-dimensional coordinate system with axes corresponding to the traveling direction, height direction, and lateral direction (i.e., directions perpendicular to the traveling direction and height direction) of the vehicle-mounted device 1, and the grid is formed by dividing the space at equal intervals along each of these directions. The downsampling size is an initial size stored in the memory 12 or the size most recently set by the downsampling size setting block 19. The downsampling block 16 supplies the generated processed point cloud data to the vehicle position estimation unit 15.

[0037] The downsampling size may be different for each of the traveling direction, height direction, and width direction of the vehicle-mounted device 1. The processing of the downsampling block 16 may also use any voxel grid filter method.

[0038] The corresponding measurement point number acquisition block 17 acquires the number of corresponding measurement points Nc from the vehicle position estimation unit 15. In this case, the vehicle position estimation unit 15 performs NDT matching for each processing time on the processed point cloud data supplied from the downsampling block 16, and outputs the number of corresponding measurement points Nc for each processing time to the corresponding measurement point number acquisition block 17. The corresponding measurement point number acquisition block 17 supplies the number of corresponding measurement points Nc to the comparison block 18.

[0039] The comparison block 18 compares the target range R indicated by the target range information 9 at each processing time. Nc is compared with the number of corresponding measurement points Nc, and the comparison result is supplied to the downsampling size setting block 19. In this case, the comparison block 18 performs, for example, The number of corresponding measurement points Nc is within the target range R Nc or The number of corresponding measurement points Nc is within the target range R Nc is within, or The number of corresponding measurement points Nc is within the target range R Nc is less than the lower limit of The result of the determination is supplied to the downsampling size setting block 19. For example, if the target range R Nc is between 600 and 800, the comparison block 18 Nc The upper limit is set to 800, and the target range is set to R Nc The above judgment is made with the lower limit of the target range R Nc is not limited to 600 to 800, and is determined depending on the hardware used, the performance of the processor, and the processing time to be completed (that is, the time interval to the next processing time).

[0040] The downsampling size setting block 19 sets the size of downsampling to be applied at the next processing time based on the comparison result of the comparison block 18. In this case, the ... Nc If the number of corresponding measurement points Nc is larger than the upper limit of the target range R, the downsampling size is increased. NcOn the other hand, the downsampling size setting block 19 determines whether the number of corresponding measurement points Nc is within the target range R Nc If it is within the range, the downsampling size is maintained (i.e., not changed).

[0041] FIG. 3 also shows a specific example of the downsampling size setting block 19 having a gain setting sub-block 191 and a multiplication sub-block 192.

[0042] The gain setting sub-block 191 sets a gain by which the current downsampling size is multiplied in the multiplication sub-block 192 based on the comparison result in the comparison block 18. In this case, the ... Nc If the number of corresponding measurement points Nc is greater than the upper limit of the target range R, the gain is increased to more than 1. Nc If the number of corresponding measurement points Nc is less than the lower limit of the target range R Nc If the gain is within the range, the gain is set to 1. In Figure 3, the gain is set as follows as an example. The number of corresponding measurement points Nc is within the target range R Nc More than the upper limit ⇒ Gain "1.1", The number of corresponding measurement points Nc is within the target range R Nc Less than the lower limit ⇒ Gain "1 / 1.1", The number of corresponding measurement points Nc is within the target range R Nc Within the range ⇒ Gain "1.0"

[0043] The multiplication sub-block 192 outputs the value obtained by multiplying the current down-sampling size by the gain set by the gain setting sub-block 191 as the down-sampling size to be used at the next processing time.

[0044] According to this configuration, the downsampling size setting block 19 determines whether the number of corresponding measurement points Nc is within the target range R NcIf the number of corresponding measurement points Nc is greater than the upper limit of the target range R, the downsampling size can be increased to change the downsampling size so that the number of measurement points in the processing point cloud data is more likely to decrease. Nc If the number of measurement points is less than the lower limit, the downsampling size can be reduced and the downsampling size can be changed so that the number of measurement points in the processing point cloud data is more likely to increase.

[0045] The gain values ​​set in the gain setting sub-block 191 are merely examples, and are set to appropriate values ​​based on experimental results, etc. Also, the downsampling size setting block 19 may add or subtract a predetermined value instead of multiplying the gain. In this case, the downsampling size setting block 19 calculates the gain value when the number of corresponding measurement points Nc is within the target range R Nc If the number of corresponding measurement points Nc is greater than the upper limit of the target range R, the downsampling size is increased by a predetermined positive value. Nc If the gain is less than the lower limit, the downsampling size is reduced by a predetermined positive value. Note that the gain and predetermined value may be set to different values ​​for each direction (travel direction, horizontal direction, and height direction).

[0046] Here, a supplementary explanation will be given regarding the appropriateness of the configuration of the downsampling processing unit 14 shown in FIG.

[0047] Generally, the larger the downsampling size, the smaller the number of data (i.e., the number of measurement points in the processing point cloud data), but the relationship between the downsampling size and the number of data is not linear. Therefore, it is not appropriate to determine the downsampling size by linear feedback control. Taking the above into consideration, in this embodiment, a target range R is set, which represents the target number of corresponding measurement points Nc as a range. Nc By using this, a dead zone is set, and the number of data and the target range R NcThe downsampling processing unit 14 is configured so that the number of corresponding measurement points Nc gradually reaches the target level by setting a constant gain regardless of the difference amount.

[0048] Next, a specific example of downsampling will be described. For simplicity of explanation, the following description will be made on the basis of voxel data VD corresponding to four voxels and point cloud data belonging to these voxels.

[0049] 4(A) to 4(C) are diagrams showing the results of a first execution of downsampling using different downsampling sizes. In FIG. 4(A), the downsampling size is set to 1x the voxel size, in FIG. 4(B), the downsampling size is set to 2 / 3x the voxel size, and in FIG. 4(C), the downsampling size is set to 1 / 2x the voxel size. In FIG. 4(A) to 4(C), the voxel data VD expressed by a normal distribution, the point cloud data before downsampling, and the processed point cloud data after downsampling are clearly shown.

[0050] The downsampling block 16 generates grids by dividing the space corresponding to the four target voxels using the downsampling size set for each case in Figures 4(A) to 4(C).The downsampling block 16 then calculates the average of the positions indicated by the measurement points of the lidar 2 for each grid, and generates the calculated average position as the measurement point of the processing point cloud data corresponding to the target grid.

[0051] Here, we consider the case where the target number of measurement points after downsampling (i.e., the number of corresponding measurement points Nc) is four. That is, we assume that the target number is four in order to achieve both sufficient position estimation accuracy and limit the calculation time. In the examples of Figures 4(A) to 4(C), the measurement points before downsampling are relatively densely packed. Therefore, in the examples of Figures 4(B) and 4(C), where the downsampling size is relatively small, the number of measurement points after downsampling becomes excessive. In other words, the matching calculation load increases, and there is a possibility that the calculation will not be completed within the specified processing time. On the other hand, in the example of Figure 4(A), where the downsampling size is the largest, the number of measurement points after downsampling is four. Therefore, in the examples of Figures 4(A) to 4(C), the downsampling size can be set to the same as the voxel size to optimize the number of measurement points after downsampling.

[0052] Figures 5(A) to 5(C) are diagrams showing the results of a second execution in which downsampling was performed using different downsampling sizes. The downsampling size in Figure 5(A) is set to 1x the voxel size, the downsampling size in Figure 5(B) is set to 2 / 3x the voxel size, and the downsampling size in Figure 5(C) is set to 1 / 2x the voxel size. We will now consider the case in which the target number of measurement points after downsampling (i.e., the number of corresponding measurement points Nc) is 4.

[0053] In this case, the voxel data and measurement points before downsampling are relatively sparse. Therefore, in the examples of Figures 5(A) and 5(B), where the downsampling size is relatively large, the number of measurement points after downsampling is too small. In other words, the small number of matching targets may result in a deterioration in position estimation accuracy. In particular, in the case of Figure 5(A), it can be seen that the rotation direction is not determined when matching voxel data and measurement points. On the other hand, in the example of Figure 5(C), where the downsampling size is the smallest, the number of measurement points after downsampling is four. Even if there is only one voxel data, it can be seen that matching is possible because the voxel data is surrounded by four measurement points. Therefore, in the examples of Figures 5(A) to 5(C), the number of measurement points after downsampling can be optimized by setting the downsampling size to half the voxel size.

[0054] Figures 6(A) to 6(C) are diagrams showing the results of a third execution in which downsampling was performed using different downsampling sizes. The downsampling size in Figure 6(A) is set to 1x the voxel size, the downsampling size in Figure 6(B) is set to 2 / 3x the voxel size, and the downsampling size in Figure 6(C) is set to 1 / 2x the voxel size. We will now consider the case where the target number of measurement points after downsampling (i.e., the number of corresponding measurement points Nc) is 4.

[0055] In this case, in the example of Figure 6(A), where the downsampling size is the largest, the number of measurement points after downsampling is too small, and in the example of Figure 6(C), where the downsampling size is the smallest, the number of measurement points after downsampling is too large. On the other hand, in the example of Figure 6(B), where the downsampling size is intermediate, the number of measurement points after downsampling is four. Therefore, in the examples of Figures 6(A) to 6(C), the number of measurement points after downsampling can be optimized by setting the downsampling size to 2 / 3 of the voxel size.

[0056] In this way, the appropriate downsampling size differs depending on the point cloud data obtained. In consideration of the above, in this embodiment, the downsampling size is adaptively changed to adjust the number of corresponding measurement points Nc to the target range R Nc The calculation time is kept within a required predetermined time, and the accuracy of the vehicle position estimation is maintained.

[0057] Here, the effect of adaptively changing the downsampling size will be described with reference to FIGS. 7(A) to 7(C).

[0058] 7(A) to 7(C) show bird's-eye views of the area around a vehicle equipped with an onboard device 1 when the downsampling size is set to 1x the voxel size. In FIGS. 7(A) to 7(C), voxels containing voxel data VD are indicated by rectangular frames, and the positions of measurement points of the LIDAR 2 obtained by one scanning cycle are indicated by dots. Here, FIG. 7(A) shows an example where there are many structures around the LIDAR 2, FIG. 7(B) shows an example where there are few structures around the LIDAR 2, and FIG. 7(C) shows an example where a phenomenon occurs in which the LIDAR light beam is blocked by another vehicle nearby (called occlusion). Note that voxels corresponding to the surface positions of structures contain corresponding voxel data VD.

[0059] As shown in Figures 7(A) to 7(C), the number of measurement points in the point cloud data acquired by the LIDAR 2 depends not only on the measurement area and field of view of the LIDAR 2 but also on the environment around the vehicle. For example, when the surrounding space is dense (see Figure 7(A)), the number of measurement points in the point cloud data is large, but when the surrounding space is sparse (see Figure 7(B)), the number of measurement points in the point cloud data is small. Furthermore, when occlusion by other vehicles or the like occurs (see Figure 7(C)), the number of corresponding measurement points Nc available for NDT scan matching decreases. In the example of Figure 7(A), the number of corresponding measurement points Nc available for NDT scan matching is sufficient, so the position estimation accuracy by NDT scan matching is high. However, in the examples of Figures 7(B) and 7(C), the number of corresponding measurement points Nc is small, so the position estimation accuracy by NDT scan matching decreases. Therefore, it is necessary to reduce the downsampling size and increase the number of measurement points Nc. If there are more surrounding structures than in Figure 7(A) and the number of measurement points Nc increases, there is a possibility that the calculation will not be completed within the specified time. In that case, it is necessary to set the downsampling size to a larger value.

[0060] Taking the above into consideration, the vehicle-mounted device 1 according to this embodiment adaptively changes the downsampling size even when the environment around the vehicle changes, thereby adjusting the number of corresponding measurement points Nc to fall within the target range R Nc This makes it possible to maintain the position estimation accuracy of NDT scan matching in an optimal manner.

[0061] (4) Position estimation based on NDT scan matching Next, the position estimation based on NDT scan matching executed by the vehicle position estimation unit 15 will be described.

[0062] FIG. 8 is a diagram showing the vehicle position to be estimated by the vehicle position estimation unit 15 in two-dimensional Cartesian coordinates. As shown in FIG. 8, the vehicle position on a plane defined on the two-dimensional Cartesian coordinates of x and y is represented by coordinates "(x, y)" and the vehicle's orientation (yaw angle) "ψ." Here, the yaw angle ψ is defined as the angle between the vehicle's traveling direction and the x-axis. The coordinates (x, y) are, for example, an absolute position corresponding to a combination of latitude and longitude, or world coordinates indicating a position with a predetermined point as the origin. The vehicle position estimation unit 15 then estimates the vehicle position using these x, y, and ψ as estimation parameters. Note that the vehicle position estimation unit 15 may also estimate the vehicle position by further estimating at least one of the vehicle's height position, pitch angle, and roll angle in a three-dimensional Cartesian coordinate system as estimation parameters in addition to x, y, and ψ.

[0063] Next, we will explain the voxel data VD used in NDT scan matching. The voxel data VD includes data in which measured point cloud data of a stationary structure in each voxel is expressed using a normal distribution.

[0064] Fig. 9 shows an example of a schematic data structure of the voxel data VD. The voxel data VD includes parameter information when expressing a point group in a voxel using a normal distribution, and in this embodiment, as shown in Fig. 9, includes a voxel ID, voxel coordinates, a mean vector, and a covariance matrix.

[0065] "Voxel coordinates" indicate the absolute three-dimensional coordinates of a reference position, such as the center position of each voxel. Each voxel is a cube that divides space into a grid, and since its shape and size are predetermined, it is possible to identify the space of each voxel using its voxel coordinates. Voxel coordinates may also be used as a voxel ID.

[0066] The "mean vector" and "covariance matrix" refer to the mean vector and covariance matrix, which are parameters when expressing the point cloud in the target voxel as a normal distribution. Note that the coordinates of an arbitrary point "i" in an arbitrary voxel "n" are X n (i)=[x n(i), y n (i), z n (i)] T and the number of points in voxel n is defined as "N n ", then the mean vector at voxel n is "μ n ” and the covariance matrix “V n " are expressed by the following formulas (1) and (2), respectively.

[0067]

number

[0068]

number

[0069] Next, an overview of NDT scan matching using voxel data VD will be explained.

[0070] Scan matching using NDT, which assumes a vehicle, estimates parameters based on the amount of movement within the road plane (here, xy coordinates) and the vehicle's orientation. P=[t x , t y , t ψ ] T Here, "t x " indicates the amount of movement in the x direction, and "t y " indicates the amount of movement in the y direction, and "t ψ " indicates the yaw angle.

[0071] In addition, the coordinates of the point cloud data output by LIDAR 2 are X L (j)=[x(j), y(j), z(j)] T Then, X L The average value of (j) "L' n " is expressed by the following equation (3).

[0072]

number

[0073]

number

[0074]

number

[0075] Then, the vehicle position estimation unit 15 calculates a comprehensive evaluation function value (also called a "score value") "E(k)" for all voxels to be matched, as shown in the following equation (6).

[0076]

number

[0077]

number

[0078] 10 is an example of functional blocks of the vehicle position estimation unit 15. As shown in FIG. 10, the vehicle position estimation unit 15 includes a dead reckoning block 21, a position prediction block 22, a coordinate transformation block 23, a point cloud data association block 24, and a position correction block 25.

[0079] The dead reckoning block 21 calculates the travel distance and change in direction from the previous time using the travel speed and angular velocity of the vehicle based on the outputs of the gyro sensor 3, the vehicle speed sensor 4, and the GPS receiver 5. The position prediction block 22 calculates the estimated vehicle position X at time k-1 calculated in the immediately preceding measurement update step. ^ Add the calculated travel distance and heading change to (k-1) to calculate the predicted vehicle position X at time k. - Calculate (k).

[0080] The coordinate transformation block 23 transforms the processed point cloud data after downsampling output from the downsampling processing unit 14 into a world coordinate system, which is the same coordinate system as the map DB 10. In this case, the coordinate transformation block 23 performs coordinate transformation of the processed point cloud data at time k, for example, based on the predicted vehicle position output by the position prediction block 22 at time k. Note that instead of performing the above-described coordinate transformation on the processed point cloud data after downsampling, the above-described coordinate transformation may be performed on the point cloud data before downsampling. In this case, the downsampling processing unit 14 generates processed point cloud data in a world coordinate system by downsampling the point cloud data in the world coordinate system after coordinate transformation. Note that the process of transforming point cloud data in a coordinate system based on a lidar installed on a vehicle into a vehicle coordinate system, and the process of transforming from the vehicle coordinate system to a world coordinate system are disclosed, for example, in International Publication WO2019 / 188745.

[0081] The point cloud data correspondence block 24 associates the processed point cloud data with the voxels by comparing the processed point cloud data in the world coordinate system output by the coordinate transformation block 23 with the voxel data VD expressed in the same world coordinate system. The position correction block 25 calculates an individual evaluation function value based on equation (5) for each voxel associated with the processed point cloud data, and calculates an estimated parameter P that maximizes the score value E(k) based on equation (6). Then, the position correction block 25 calculates the predicted vehicle position X output by the position prediction block 22 based on equation (7). - (k), the estimated vehicle position X ^ Calculate (k).

[0082] Here, a specific procedure for associating measurement points with voxel data VD will be explained further using a simple example.

[0083] 11 shows the positional relationship between voxels "Vo1" to "Vo6" in which voxel data VD exists on a two-dimensional xy plane in the world coordinate system and measurement points 61 to 65 that indicate positions near these voxels. For ease of explanation, it is assumed here that the z coordinate of the center positions of voxels Vo1 to Vo6 in the world coordinate system is the same as the z coordinate of measurement points 61 to 65 in the world coordinate system.

[0084] First, the coordinate conversion block 23 converts the point cloud data including the measurement points 61 to 65 into the world coordinate system. Then, the point cloud data association block 24 rounds off fractions of the measurement points 61 to 65 in the world coordinate system. In the example of Fig. 11, since the size of each cubic voxel is 1m, the point cloud data association block 24 rounds off the decimal points of the x, y, and z coordinates of each of the measurement points 61 to 65.

[0085] Next, the point cloud data association block 24 determines the voxels corresponding to each of the measurement points 61 to 65 by comparing the voxel data VD corresponding to the voxels Vo1 to Vo6 with the coordinates of each of the measurement points 61 to 65. In the example of FIG. 11 , the (x, y) coordinates of the measurement point 61 become (2, 1) after the above-mentioned rounding, so the point cloud data association block 24 associates the measurement point 61 with the voxel Vo1. Similarly, the (x, y) coordinates of the measurement points 62 and 63 become (3, 2) after the above-mentioned rounding, so the point cloud data association block 24 associates the measurement points 62 and 63 with the voxel Vo5. Furthermore, the (x, y) coordinates of the measurement point 64 become (2, 3) after the above-mentioned rounding, so the point cloud data association block 24 associates the measurement point 64 with the voxel Vo6. On the other hand, because the (x, y) coordinates of measurement point 65 become (4, 1) after the above-mentioned rounding, the point cloud data association block 24 determines that there is no voxel data VD corresponding to measurement point 65. Thereafter, the position correction block 25 estimates the estimation parameters P using the measurement point and voxel data VD associated by the point cloud data association block 24. Note that in this example, the number of measurement points after downsampling is five, but the number of corresponding measurement points Nc is four. As mentioned above, the number of corresponding measurement points Nc becomes smaller when there are few structures around the road and the space is sparse, or when there is an occlusion due to the presence of another vehicle near the host vehicle.

[0086] (5) Processing flow Fig. 12 is an example of a flowchart showing the procedure of processing related to vehicle position estimation executed by the controller 13 of the vehicle-mounted device 1. The controller 13 starts the processing of the flowchart in Fig. 12 when it becomes necessary to estimate the vehicle position, for example, when the power is turned on.

[0087] First, the controller 13 sets the down-sampling size to an initial value (step S11). Here, as an example, the down-sampling size is set to 1 m in each of the traveling direction, lateral direction, and vertical direction. Next, the controller 13 determines a predicted vehicle position based on the positioning result of the GPS receiver 5 (step S12).

[0088] Then, the controller 13 determines whether or not the point cloud data of the LIDAR 2 has been acquired (step S13). If the point cloud data of the LIDAR 2 cannot be acquired (step S13; No), the controller 13 continues to perform the determination of step S13. Furthermore, during the period when the point cloud data cannot be acquired, the controller 13 continues to determine the predicted vehicle position based on the positioning result of the GPS receiver 5. Note that the controller 13 may determine the predicted vehicle position based on the output of any sensor other than the point cloud data, not limited to the positioning result of the GPS receiver 5 alone.

[0089] Then, when the controller 13 has acquired the point cloud data of the LIDAR 2 (step S13; Yes), it performs downsampling on the point cloud data for one cycle of scanning by the LIDAR 2 at the current processing time using the current setting value of the downsampling size (step S14). In this case, the setting value of the downsampling size corresponds to the initial value determined in step S11 or the updated value determined in the immediately preceding steps S19 to S22.

[0090] Then, the controller 13 obtains the travel distance and change in orientation from the previous time using the vehicle's travel speed and angular velocity based on the outputs of the gyro sensor 3, the vehicle speed sensor 4, etc. As a result, the position prediction block 22 calculates the predicted vehicle position (which may include an attitude angle) at the current processing time from the estimated vehicle position (which may include an attitude angle such as a yaw angle) obtained one time before (the immediately previous processing time) (step S15). Note that if the estimated vehicle position one time before has not been calculated (i.e., if step S23 has not been executed), the controller 13 may perform the processing of step S15 by regarding the predicted vehicle position determined in step S12 as the estimated vehicle position.

[0091] Next, the controller 13 executes an NDT matching process (step S16). In this case, the controller 13 performs a process of converting the processing point cloud data into data in a world coordinate system, and a process of associating the processing point cloud data converted into the world coordinate system with voxels in which the voxel data VD exists. The controller 13 may also perform a process of calculating an estimated vehicle position (including an attitude angle such as a yaw angle) at the current processing time based on the NDT matching.

[0092] Furthermore, the controller 13 calculates the number of corresponding measurement points Nc based on the result of associating the processing point cloud data with the voxel data VD in the NDT matching process (step S17). Then, the controller 13 calculates the number of corresponding measurement points Nc within the target range R indicated by the target range information 9. Nc (Step S18). Here, as an example, the target range R Nc The explanation will be given assuming that the value is 600 to 800.

[0093] And the number of corresponding measurement points Nc is within the target range R Nc If the downsampling size is less than 600, which corresponds to the lower limit of 1, the controller 13 changes the set value of the downsampling size to 1 / 1.1 (step S19). Note that "1 / 1.1" is just an example, and the controller 13 may change the set value by subtracting any factor less than 1 or a value greater than 0. If a lower limit is set for the downsampling size, the controller 13 limits the change to the set value of the downsampling size so that it does not fall below the lower limit (step S20).

[0094] In addition, the number of corresponding measurement points Nc is within the target range R NcIf the number of corresponding measurement points Nc is greater than 800, which corresponds to the upper limit of the target range R, the controller 13 changes the set value of the down-sampling size to 1.1 (step S21). Note that "1.1" is just an example, and the controller 13 may change the set value by adding any magnification greater than 1 or a value greater than 0. If an upper limit is set for the down-sampling size, the controller 13 limits the change in the set value of the down-sampling size so that the upper limit is not exceeded (step S22). Note that the controller 13 also changes the set value of the down-sampling size to 1.1 (step S22) if the number of corresponding measurement points Nc is greater than 800, which corresponds to the upper limit of the target range R. Nc If it belongs to, the downsampling size setting is maintained.

[0095] Then, based on the result of the NDT matching, the controller 13 calculates an estimated vehicle position (including attitude angles such as yaw angle) at the current processing time (step S23). The controller 13 then determines whether or not the vehicle position estimation process should be ended (step S24). If the controller 13 determines that the vehicle position estimation process should be ended (step S24; Yes), the controller 13 ends the processing of the flowchart. On the other hand, if the controller 13 determines that the vehicle position estimation process should be continued (step S24; No), the controller 13 returns the processing to step S13 and estimates the vehicle position at the next processing time.

[0096] (6) Variations The following describes preferred modifications of the above-described embodiment. The following modifications may be applied to these embodiments in combination.

[0097] (Variation 1) The vehicle-mounted device 1 may set the downsampling size to a fixed value in a direction (also called a "low-accuracy direction") in which the position estimation accuracy is relatively poor among the directions constituting the coordinate axes of the three-dimensional coordinate system that is the space in which downsampling is performed, and adaptively change the downsampling size in other directions based on the above-described embodiment. The low-accuracy direction is, for example, the traveling direction. The low-accuracy direction may also be the horizontal direction or the vertical direction.

[0098] In this case, for example, the low-accuracy direction is empirically identified before execution of the vehicle position estimation process based on the specifications of the LIDAR 2 or the driving assistance system used, and a fixed value of the downsampling size in the low-accuracy direction is stored in the memory 12. Then, the vehicle-mounted device 1 sets the downsampling size in the low-accuracy direction to the fixed value stored in the memory 12, and changes the downsampling sizes in the other directions in steps S19 to S22 of the flowchart in Fig. 12. In this way, the vehicle-mounted device 1 can ensure the resolution of the processed point cloud data in the low-accuracy direction and reliably suppress a decrease in the position estimation accuracy in the low-accuracy direction.

[0099] (Variation 2) The vehicle-mounted device 1 may set the downsampling size in the low-precision direction to be less than one time the downsampling size in the other directions.

[0100] For example, the vehicle-mounted device 1 determines the down-sampling size for each direction based on the size (also called the "reference size") determined based on the number of corresponding measurement points Nc and a ratio set in advance for each direction depending on accuracy, importance, etc. For example, the vehicle-mounted device 1 sets the down-sampling size for the traveling direction to 1 / 2 the reference size determined in step S11 or steps S19 to S22 of the flowchart in FIG. 12. On the other hand, the vehicle-mounted device 1 sets the down-sampling size for the horizontal direction to twice the reference size, and the down-sampling size for the height direction to be the same as the reference size (same size). In this way, the vehicle-mounted device 1 can always increase the relative resolution of the processed point cloud data for directions other than the low-accuracy direction.

[0101] (Variation 3) The vehicle-mounted device 1 may measure the processing time (length of processing time) of the vehicle position estimation process (including downsampling) executed at each processing time, and set the downsampling size according to the measured processing time.

[0102] For example, when the execution cycle of the vehicle position estimation process is 83.3 ms, the vehicle-mounted device 1 reduces the downsampling size by a predetermined value or a predetermined rate when the processing time is less than 30 ms. On the other hand, when the processing time is longer than 70 ms, the vehicle-mounted device 1 increases the downsampling size by a predetermined value or a predetermined rate. Furthermore, when the processing time is between 30 ms and 70 ms, the vehicle-mounted device 1 maintains the downsampling size setting. Note that the vehicle-mounted device 1 may change the downsampling size setting based on this modification after, for example, executing step S23 in FIG. 12, or may execute it immediately before or after steps S18 to S22.

[0103] According to this modification, the vehicle-mounted device 1 can suitably set the downsampling size so that the processing time falls within an appropriate range for the execution period.

[0104] (Variation 4) The configuration of the driving assistance system shown in Fig. 1 is one example, and the configuration of a driving assistance system to which the present invention can be applied is not limited to the configuration shown in Fig. 1. For example, instead of having an on-board device 1, the driving assistance system may have an electronic control unit of the vehicle that executes the processing of the downsampling processing unit 14 and the vehicle position estimation unit 15 of the on-board device 1. In this case, the map DB 10 is stored in, for example, a storage unit in the vehicle or a server device that communicates data with the vehicle, and the electronic control unit of the vehicle references this map DB 10 to execute downsampling and vehicle position estimation based on NDT scan matching.

[0105] (Variation 5) The voxel data VD is not limited to a data structure including a mean vector and a covariance matrix as shown in Fig. 9. For example, the voxel data VD may include point cloud data measured by a measurement and maintenance vehicle that is used to calculate the mean vector and the covariance matrix.

[0106] (7) Considerations based on the experimental results Next, we will consider the experimental results for the above-mentioned embodiment (including modifications). The applicant drove a vehicle equipped with two front and two rear LIDARs, each with a horizontal field of view of 60 degrees and an operating frequency of 12 Hz (83.3 ms period), along a certain road, and performed position estimation based on NDT by matching voxel data (ND map) created in advance for the road with LIDAR point cloud data obtained during driving. In addition, for accuracy evaluation, RTK-GPS positioning results were used as the correct position data.

[0107] Figures 13(A) to 13(F) and 14(A) to 14(E) show the results of downsampling to a cubic grid size of 0.5 m on a side and estimating vehicle position using NDT processing. For the RTK-GPS positioning results, Figure 13(A) shows the error in the direction of travel, Figure 13(B) shows the error in the lateral direction, Figure 13(C) shows the error in the vertical direction, and Figure 13(D) shows the error in the yaw angle. Figure 13(E) shows the number of measurement points in the point cloud data before downsampling, and Figure 13(F) shows the downsampling size. Figure 14(A) shows the number of measurement points in the processed point cloud data after downsampling, Figure 14(B) shows the number of corresponding measurement points Nc, Figure 14(C) shows the ratio of the number of corresponding measurement points Nc to the number of measurement points in the processed point cloud data (also called the "correspondence rate"). Figure 14(D) shows the score value E(k), and Figure 14(E) shows the processing time. In FIG. 14(E), the period of the lidar (83.3 ms) is indicated by a horizontal line.

[0108] Here, as shown in Fig. 14(E), there is a period in which the processing time exceeds the lidar period of 83.3 ms. During this period, the number of measurement points before downsampling shown in Fig. 13(E), the number of measurement points after downsampling shown in Fig. 14(A), and the number of corresponding measurement points Nc shown in Fig. 14(B) are all large. As such, it can be seen that there is a positive correlation between the number of point cloud data and processing time.

[0109] 13(A), the error in the direction of travel momentarily increases around 170 s. In addition to the fact that the area traveled at this time was an area with few features in the direction of travel, the processing time exceeded the lidar cycle, and the processing was not completed in time, which combined to cause the estimated position to deviate.

[0110] As such, processing must be performed in time with the LIDAR period, and if downsampling is set to a fixed size, this problem may not be solved.

[0111] Figures 15(A) to 15(F) and 16(A) to 16(E) show the results of downsampling using a cubic grid size of 2.0 m on a side and estimating vehicle position using NDT processing. For the RTK-GPS positioning results, Figure 15(A) shows the error in the direction of travel, Figure 15(B) shows the error in the lateral direction, Figure 15(C) shows the error in the vertical direction, and Figure 15(D) shows the error in the yaw angle. Figure 15(E) shows the number of measurement points in the point cloud data before downsampling, and Figure 15(F) shows the downsampling size. Figure 16(A) shows the number of measurement points in the processed point cloud data after downsampling, Figure 16(B) shows the number of corresponding measurement points Nc, Figure 16(C) shows the correspondence rate, Figure 16(D) shows the score value E(k), and Figure 16(E) shows the processing time. In FIG. 16(E), the period of the lidar (83.3 ms) is indicated by a horizontal line.

[0112] In this case, compared to the examples in Figures 13(A) to 13(F) and Figures 14(A) to 14(E), increasing the downsampling size reduces the number of corresponding measurement points Nc shown in Figure 16(B) and the processing time shown in Figure 16(E). As a result, the processing time is within the lidar period of 83.3 ms at all times. On the other hand, the error in the direction of travel shown in Figure 15(A) increases (especially from 250 s to 280 s). This is presumably due to a decrease in the resolution of the processing point cloud data, resulting in a decrease in the number of measurement points after downsampling. Note that if the number of measurement points in the processing point cloud data is too small, there will be an insufficient number of measurement points to match in the NDT process, which may reduce stability.

[0113] Thus, even if the downsampling size is set to a large fixed size, it is not possible to maintain the accuracy of estimating the vehicle position while reducing the processing time.

[0114] Figures 17(A) to 17(F) and 18(A) to 18(E) show the results of vehicle position estimation using downsampling and NDT processing according to the processing procedure shown in Figure 12 according to this embodiment. For the RTK-GPS positioning results, Figure 17(A) shows the heading error, Figure 17(B) shows the lateral error, Figure 17(C) shows the vertical error, and Figure 17(D) shows the yaw angle error. Figure 17(E) shows the number of measurement points in the point cloud data before downsampling, and Figure 17(F) shows the downsampling size. Furthermore, Figure 18(A) shows the number of measurement points in the processed point cloud data after downsampling, Figure 18(B) shows the number of corresponding measurement points Nc, Figure 18(C) shows the correspondence rate, Figure 18(D) shows the score value E(k), and Figure 18(E) shows the processing time. In FIG. 18(E), the period of the lidar (83.3 ms) is indicated by a horizontal line.

[0115] In this case, the downsampling size is dynamically changed as shown in Fig. 17(F). As a result, the processing time shown in Fig. 18(E) is always less than the lidar period of 83.3 ms, and the heading error shown in Fig. 17(A) is also effectively suppressed.

[0116] In this way, by adaptively changing the downsampling size based on this embodiment, it is possible to maintain the accuracy of estimating the vehicle position while reducing the processing time.

[0117] Figures 19(A) to 19(G) show the results of estimating the vehicle position by downsampling and NDT processing in accordance with the processing procedure shown in Figure 12 according to this embodiment when occlusion by another vehicle occurs. Figures 20(A) to 20(G) show the results when, under the same circumstances as Figures 19(A) to 19(G), the downsampling size is determined so that the number of measurement points after downsampling falls within the range of 600 to 800, instead of the number of corresponding measurement points Nc. Here, Figures 19(A) and 20(A) show the number of measurement points before downsampling, Figures 19(B) and 20(B) show the downsampling size, Figures 19(C) and 20(C) show the number of measurement points after downsampling, Figures 19(D) and 20(D) show the number of corresponding measurement points Nc, Figures 19(E) and 20(E) show the correspondence rate, Figures 19(F) and 20(F) show the score value E(k), and Figures 19(G) and 20(G) show the processing time.

[0118] In these examples, occlusion by other vehicles occurs for a while after 180 seconds. If the downsampling size is controlled based on the number of measurement points before they are associated with the voxel data VD, as in the examples of Figures 20(A) to 20(G), the number of corresponding measurement points Nc becomes too small when occlusion occurs. As a result, as shown in Figure 20(D), the number of corresponding measurement points Nc decreases during the period when occlusion occurs. On the other hand, when the downsampling size according to this embodiment is controlled, the fluctuation in the number of corresponding measurement points Nc is small even during the period when occlusion occurs, as shown in Figure 19(D). Therefore, the examples of Figures 20(A) to 20(G) result in lower stability of position estimation compared to the examples of Figures 19(A) to 19(G). This example shows that when there are many measurement points that cannot be associated with voxel data based on stationary structures due to the occurrence of occlusion, it is not effective to control the downsampling so that the measurement points after downsampling fall within the target range.

[0119] Figures 21(A) to 21(F) and 22(A) to 22(E) show the results of vehicle position estimation using NDT processing, with the downsampling size in the traveling direction fixed at 0.5 m and the downsampling sizes in the horizontal and vertical directions adaptively changed based on Modification 1. Specifically, for the RTK-GPS positioning results, Figure 21(A) shows the traveling direction error, Figure 21(B) shows the horizontal error, Figure 21(C) shows the vertical error, and Figure 21(D) shows the yaw angle error. Figure 21(E) shows the number of measurement points in the point cloud data before downsampling, and Figure 21(F) shows the downsampling size. Furthermore, Figure 22(A) shows the number of measurement points in the processed point cloud data after downsampling, Figure 22(B) shows the number of corresponding measurement points Nc, Figure 22(C) shows the correspondence rate, Figure 22(D) shows the score value E(k), and Figure 22(E) shows the processing time. In FIG. 22(E), the period of the lidar (83.3 ms) is indicated by a horizontal line.

[0120] In this example, there are many guardrails, road shoulders, and medians on both sides of the road, and considering that lateral errors are more accurate than errors in the direction of travel, the direction of travel is considered to be the low-accuracy direction, and the downsampling size in the direction of travel is fixed to 0.5 m. Furthermore, by fixing the downsampling size in the direction of travel to 0.5 m, as shown in FIG. 21(F), the downsampling sizes in the lateral and vertical directions are larger than when the downsampling size is adaptively changed, including the direction of travel (see FIG. 17(F)). Meanwhile, in Modification 1, by controlling the number of corresponding measurement points Nc shown in FIG. 22(B) to be maintained between 600 and 800, the processing time shown in FIG. 22(E) is also suitably kept within the lidar period.

[0121] As a result, when Modification 1 is executed, the lateral error shown in Fig. 21(B) remains almost unchanged, and the error in the direction of travel shown in Fig. 21(A) (particularly around 170s) is reduced. In this way, even when controlling the downsampling size based on Modification 1, it is possible to favorably maintain the accuracy of vehicle position estimation while reducing processing time.

[0122] As described above, the controller 13 of the vehicle-mounted device 1 according to this embodiment acquires point cloud data output by the LIDAR 2. The controller 13 then generates processed point cloud data by downsampling the point cloud data. The controller 13 then matches the processed point cloud data with voxel data VD representing the position of an object for each voxel, which is a unit area, to associate the measurement points constituting the processed point cloud data with each voxel. The controller 13 then changes the size of the downsampling to be performed next based on the number of corresponding measurement points Nc, which is the number of measurement points in the processed point cloud data associated with a voxel. This allows the vehicle-mounted device 1 to adaptively determine the downsampling size, thereby achieving both high accuracy and low processing time for self-localization estimation.

[0123] In the above-described embodiments, the program can be stored using various types of non-transitory computer-readable media and supplied to a controller or the like that is a computer. Non-transitory computer-readable media include various types of tangible storage media. Examples of non-transitory computer-readable media include magnetic storage media (e.g., flexible disks, magnetic tapes, hard disk drives), magneto-optical storage media (e.g., magneto-optical disks), CD-ROMs (Read Only Memory), CD-Rs, CD-R / Ws, and semiconductor memories (e.g., mask ROMs, PROMs (Programmable ROMs), EPROMs (Erasable PROMs), flash ROMs, and RAMs (Random Access Memory)).

[0124] Although the present invention has been described above with reference to the examples, the present invention is not limited to the above examples. Various modifications within the scope of the present invention that would be understood by those skilled in the art can be made to the configuration and details of the present invention. In other words, the present invention naturally includes various modifications and alterations that would be possible for those skilled in the art in accordance with the entire disclosure, including the claims, and the technical ideas. Furthermore, the disclosures of the above-cited patent documents and other documents are incorporated herein by reference. [Explanation of symbols]

[0125] 1 Onboard device 2 Rider 3 Gyro sensor 4 Vehicle speed sensor 5 GPS receiver 10 Map DB

Claims

1. an acquisition means for acquiring point cloud data output by a measurement device; a downsampling processing means for generating processed point cloud data by downsampling the point cloud data; a correlation means for correlating the processing point cloud data with voxel data representing the position of an object for each voxel, which is a unit area, to correlate the measurement points constituting the processing point cloud data with each of the voxels, The downsampling processing means changes the size of the downsampling to be performed next based on the number of corresponding measurement points, which is the number of measurement points associated with the voxel among the measurement points.

2. 2. The information processing device according to claim 1, wherein the downsampling processing means fixes the size in a first direction among directions in which a space is divided in the downsampling, and changes the size in directions other than the first direction based on the number of corresponding measurement points.

3. The information processing device according to claim 2 , wherein the first direction is a traveling direction of the measuring device.

4. 4. The information processing apparatus according to claim 1, wherein the downsampling processing means changes the size based on a comparison between the number of corresponding measurement points and a target range for the number of corresponding measurement points.

5. The downsampling processing means If the number of corresponding measurement points is greater than the upper limit of the target range, the size is increased by a predetermined rate or a predetermined value; If the number of corresponding measurement points is less than the lower limit of the target range, the size is reduced by a predetermined rate or a predetermined value; If the number of corresponding measurement points is within the target range, maintain the size. The information processing device according to claim 4 .

6. The information processing device according to any one of claims 1 to 5, wherein the downsampling processing means determines a reference size for the downsampling to be performed next based on the number of corresponding measurement points, and determines the size for each direction based on the reference size and a ratio set for each direction in which space is divided in the downsampling.

7. The information processing device according to any one of claims 1 to 6, further comprising a position estimation means for estimating the position of a moving body equipped with the measurement device based on a comparison result between voxel data of the voxels associated by the association means and measurement points associated with the voxels.

8. The information processing apparatus according to claim 7 , wherein the downsampling processing means changes the size based on processing times of the downsampling processing means and the position estimation means.

9. The information processing apparatus according to claim 8 , wherein the downsampling processing means changes the size based on a comparison between the processing time and a target range of the processing time.

10. A computer-implemented control method comprising: Acquire the point cloud data output by the measuring device, generating processed point cloud data by downsampling the point cloud data; By comparing the processing point cloud data with voxel data representing the position of an object for each voxel, which is a unit area, the measurement points constituting the processing point cloud data are associated with each of the voxels; A control method for changing the size of the downsampling to be performed next based on the number of corresponding measurement points, which is the number of measurement points among the measurement points that are associated with the voxel.

11. Acquire the point cloud data output by the measuring device, generating processed point cloud data by downsampling the point cloud data; By comparing the processing point cloud data with voxel data representing the position of an object for each voxel, which is a unit area, the measurement points constituting the processing point cloud data are associated with each of the voxels; A program that causes a computer to execute a process of changing the size of the downsampling to be executed next, based on the number of corresponding measurement points, which is the number of measurement points among the measurement points that are associated with the voxel.

12. A storage medium storing the program according to claim 11.

Images