Map data creation device

Abstract

To upgrade the precision in self-position estimation.SOLUTION: A controller acquires map image data obtained by imaging a road surface using a camera mounted in a vehicle. The controller links position information, which includes coordinates of a point where the map image data is acquired, to the map image data. The controller evaluates the reliability of the map image data that is the reliability concerning the precision in self-position estimation.SELECTED DRAWING: Figure 6

Description

[Technical Field]

[0001] The present disclosure relates to a map data creation device. [Background technology]

[0002] The autonomous vehicle disclosed in Patent Document 1 includes a camera, a storage device, and a control device. The camera is positioned to capture images of the road surface. The storage device stores map data. The map data is data that links map image data, which is a pre-image of the road surface, with location information. The control device acquires the image data from the camera. The control device estimates its own location by matching the image data with the map image data. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2016-166853 Summary of the Invention [Problem to be solved by the invention]

[0004] The map image data may contain low-quality image data that may make matching difficult or result in erroneous matching when performing self-location estimation. [Means for solving the problem]

[0005] A map data creation device that solves the above problem is a map data creation device that creates map data used by an autonomous vehicle when estimating its own position, and that is used to compare with image data acquired from an imaging device equipped in the autonomous vehicle.The map data creation device is equipped with a control device that acquires map image data obtained by imaging the road surface with an imaging device equipped in the vehicle, links location information including the coordinates of the point where the map image data was acquired to the map image data, and evaluates the reliability of the map image data, which is the reliability related to the accuracy of the self-position estimation.

[0006] The control device evaluates the reliability of the map image data when creating the map data. Since the reliability is related to the accuracy of self-location estimation, the accuracy of self-location estimation can be improved by using this reliability as an index.

[0007] In the above-described map data creation device, the control device may reduce the reliability of the map image data when, among the pixels constituting the map image data, the total number of pixels whose brightness is less than a low brightness threshold and pixels whose brightness is equal to or greater than a high brightness threshold is equal to or greater than a pixel threshold.

[0008] In the map data creation device, the control device may reduce the reliability of the map image data when the number of feature points included in the map image data is less than a feature point threshold.

[0009] In the map data creation device, the control device may reduce the reliability of the map image data when feature points included in the map image data are unevenly distributed. In the map data creation device, the control device may perform matching between the plurality of map image data, and if the number of matching feature points is equal to or greater than a first matching threshold, reduce the reliability of the map image data.

[0010] In the above-mentioned map data creation device, the control device may match the map image data with sample image data whose reliability evaluation value is less than a threshold, and if the number of matching feature points is equal to or greater than a second matching threshold, reduce the reliability of the map image data. [Effects of the Invention]

[0011] According to the present invention, it is possible to improve the accuracy of self-location estimation. [Brief explanation of the drawings]

[0012] [Figure 1] FIG. 1 is a side view of an autonomous vehicle. [Figure 2] FIG. 2 is a schematic configuration diagram of the autonomous vehicle of FIG. 1. [Figure 3] 3 is a flowchart showing self-position estimation executed by the control device of FIG. 2. [Figure 4] 3 is a flowchart showing map data creation control executed by the control device of FIG. 2; [Figure 5] 3 is a flowchart showing a map creation process performed by the control device of FIG. 2. [Figure 6] 3 is a flowchart showing a reliability evaluation process performed by the control device of FIG. 2. DETAILED DESCRIPTION OF THE INVENTION

[0013] An embodiment of a map data creation device will be described below. As shown in FIG. 1, the vehicle 10 includes a vehicle body 11, drive wheels 21, and steering wheels 31. The vehicle 10 may be a passenger car or an industrial vehicle. Industrial vehicles include forklifts, towing tractors, and automated guided vehicles. The vehicle 10 of this embodiment is an autonomous vehicle capable of autonomous driving.

[0014] As shown in FIG. 2, the vehicle 10 includes a traction motor 22, a traction motor driver 23, a steering motor 32, and a steering motor driver 33. The traction motor 22 is a motor for rotating the drive wheels 21. The traction motor driver 23 drives the traction motor 22. The drive of the traction motor 22 rotates the drive wheels 21, causing the vehicle 10 to travel. The steering motor 32 is a motor for steering the steered wheels 31. The steering motor driver 33 drives the steering motor 32. The drive of the steering motor 32 steers the steered wheels 31, causing the vehicle 10 to turn.

[0015] The vehicle 10 includes a control device 12. The control device 12 includes a processor 13 and a storage unit 14. Examples of the processor 13 include a central processing unit (CPU), a graphics processing unit (GPU), and a digital signal processor (DSP). The storage unit 14 includes a random access memory (RAM) and a read-only memory (ROM). The storage unit 14 stores program code or instructions configured to cause the processor 13 to execute processes. The storage unit 14, i.e., a computer-readable medium, includes any available medium accessible by a general-purpose or special-purpose computer. The control device 12 may be configured with hardware circuits such as an application-specific integrated circuit (ASIC) or a field-programmable gate array (FPGA). The control device 12, which is a processing circuit, may include one or more processors operating according to a computer program, one or more hardware circuits such as an ASIC or an FPGA, or a combination thereof.

[0016] The vehicle 10 is equipped with a camera 41. The camera 41 is an imaging device. The camera 41 is a monocular camera. The camera 41 is a digital camera. The camera 41 is equipped with an imaging element. Examples of the imaging element include a CCD image sensor (Charge Coupled Device image sensor) and a CMOS image sensor (Complementary Metal Oxide Semiconductor image sensor). Examples of the camera 41 include an RGB camera, an infrared camera, a grayscale camera, and a visible light camera.

[0017] The camera 41 captures images at a predetermined frame rate and generates image data. This image data is digital data of the images captured by the camera 41. 1, the camera 41 is disposed so as to capture an image of the road surface Sr. The camera 41 generates image data representing an image of the road surface Sr. The camera 41 is provided at the bottom of the vehicle body 11 and faces vertically.

[0018] As shown in FIG. 2, the vehicle 10 is equipped with a lighting device 51. The lighting device 51 is arranged to illuminate the road surface Sr. More specifically, the lighting device 51 illuminates the range of the road surface Sr that is captured by the camera 41. In this embodiment, the lighting device 51 is provided at the bottom of the vehicle body 11 and faces vertically. For example, a light-emitting diode can be used as the lighting device 51. The lighting device 51 illuminates the road surface Sr, thereby suppressing differences in the brightness of the road surface Sr that may occur depending on the surrounding environment of the vehicle 10.

[0019] The vehicle 10 is equipped with a positioning device 61. The positioning device 61 is equipped with a satellite navigation device 62 and an inertial measurement device 63. The satellite navigation device 62 receives satellite signals transmitted from GNSS (Global Navigation Satellite System) satellites. The satellite navigation device 62 measures its position using the satellite signals. The inertial measurement device 63 includes a gyro sensor and an acceleration sensor.

[0020] The vehicle 10 includes an auxiliary storage device 71. The auxiliary storage device 71 stores information that can be read by the control device 12. Examples of the auxiliary storage device 71 include a hard disk drive, a solid state drive, and a flash memory.

[0021] The auxiliary storage device 71 stores map data M1. The map data M1 is a combination of map image data of a road surface Sr captured in advance and position information. The range in which the vehicle 10 travels is determined in advance. The position information includes coordinates and attitude. The coordinates are coordinates of a map coordinate system, which is a coordinate system that represents an absolute position. The map coordinate system may be a Cartesian coordinate system or a geographic coordinate system. The map coordinate system has an X axis and a Y axis. The X axis and the Y axis are orthogonal to each other. The X axis and the Y axis are coordinate systems that represent the horizontal direction. The attitude is information that indicates the inclination of the vehicle 10 with respect to the coordinate axes of the map coordinate system. In this embodiment, the attitude is information that indicates the inclination with respect to the X axis.

[0022] All map image data included in the map data M1 is assigned a reliability evaluation value. The reliability evaluation value may be expressed as a binary value or as a score. When the reliability evaluation value is expressed as a binary value, the binary value indicates whether the map image data is low-quality image data or not. When the reliability evaluation value is expressed as a score, the higher the reliability of the map image data, the higher the score. In this embodiment, the reliability evaluation value is expressed as a binary value. That is, the reliability evaluation value indicates whether the map image data is low-quality image data or not. Map image data that is different from low-quality image data will be referred to as normal-quality image data as appropriate. Normal-quality image data has a higher reliability than low-quality image data. It can be said that the reliability evaluation value indicates low-quality image data and normal-quality image data.

[0023] The vehicle 10 is equipped with a wireless communication device 81. The wireless communication device 81 is a communication device capable of communicating using any wireless communication method, such as a wireless LAN (Local Area Network), ZigBee (registered trademark), LPWA (Low Power Wide Area), or a mobile communication system. The wireless communication device 81 is capable of transmitting and receiving wireless signals. This wireless signal includes position information detected by a total station. The total station is a device that detects position information of the vehicle 10. Via the wireless communication device 81, the control device 12 can acquire the position information detected by the total station.

[0024] <Self-location estimation> The control device 12 performs self-location estimation. The self-location estimation is performed repeatedly at a predetermined control cycle. The self-location estimation is performed when the vehicle 10 travels autonomously. For example, the vehicle 10 generates a travel route to a target point. Then, the vehicle 10 travels along the travel route while recognizing its own position through self-location estimation.

[0025] As shown in FIG. 3, the control device 12 acquires image data from the camera 41 in step S1. Next, in step S2, the control device 12 excludes map image data included in the map data M1 whose reliability evaluation value is less than a threshold value. The threshold value is set so that low-quality image data is excluded.

[0026] Next, in step S3, the control device 12 matches the image data with the map image data. Since low-quality image data has been excluded in step S2, the matching in step S3 is performed between the image data acquired in step S1 and the map image data from the map data M1 excluding the low-quality image data.

[0027] The control device 12 extracts feature points from the image data. The control device 12 describes the feature amounts of the feature points. Examples of feature amounts include feature vectors and brightness values. The control device 12 also extracts feature points and describes the feature amounts using map image data. The control device 12 compares the feature points and feature amounts obtained from the image data with the feature points and feature amounts obtained from the map image data, and searches for pairs of feature points with similar feature amounts. The control device 12 identifies map image data corresponding to the image data based on the feature point pairs. For example, the control device 12 identifies map image data in which feature point pairs are concentrated as map image data corresponding to the image data. The above-mentioned matching can be performed using a feature amount descriptor. Examples of the feature amount descriptor include ORB (Oriented Fast and Rotated BRIEF), SIFT (Scale-Invariant Feature Transform), and SURF (Speeded Up Robust Features).

[0028] Next, in step S4, the control device 12 estimates its own position based on map image data corresponding to the image data. The own position includes the coordinates of the vehicle 10 in a map coordinate system and the attitude of the vehicle 10. The control device 12 calculates the relative position between the map image data and the image data, and the relative angle between the map image data and the image data. The relative position between the map image data and the image data is the amount of deviation between the image data and the map image data. The relative angle between the image data and the map image data is the angle of deviation between the image data and the map image data. The image data and the map image data often do not match perfectly. This is because the position and attitude of the vehicle 10 rarely match perfectly between the time when the map image data is acquired and the time when the image data is acquired. For this reason, the image data often only partially matches the map image data. If the position of the vehicle 10 is different between the time when the map image data is acquired and the time when the image data is acquired, the difference in the position of the vehicle 10 causes a deviation between the position of the road surface Sr shown in the map image data and the position of the road surface Sr shown in the image data. This amount of deviation is the relative position between the map image data and the image data. The amount of deviation can be determined from the positional relationship between feature points in the map image data and feature points in the image data. Similarly, the image data is obtained by rotating the map image data due to the difference in the attitude of the vehicle 10 between the time when the map image data was acquired and the time when the image data was acquired. The angle of deviation caused by this rotation is the relative angle between the image data and the map image data. The control device 12 estimates its own position based on the position information, relative position, and relative angle associated with the map image data corresponding to the image data. The control device 12 shifts the coordinates associated with the map image data by the coordinates corresponding to the relative position. The control device 12 shifts the attitude associated with the map image data by the relative angle. The control device 12 determines the coordinates and attitude in the map coordinate system obtained as a result as its own position. After completing the processing of step S4, the control device 12 ends the self-position estimation.

[0029] As described above, the map data M1 is used when the vehicle 10 estimates its own position. The map data M1 is used for comparison with image data acquired from the camera 41 provided in the vehicle 10.

[0030] <Map data creation control> The map data creation control will be described. The map data creation control is control performed by the control device 12 when creating the map data M1. The map data creation control is performed while the vehicle 10 is traveling. When creating the map data M1, the vehicle 10 may be driven by a human or may be driven under the control of the control device 12.

[0031] 4, in step S11, the control device 12 performs a map creation process. The map creation process will be described in detail below. 5, in step S21, the control device 12 acquires position information. The position information can be acquired, for example, from at least one of the positioning device 61 and the wireless communication device 81. The position information that can be acquired from the positioning device 61 can be calculated from at least one of the longitude and latitude acquired using the satellite navigation device 62 and the self-movement amount calculated using the inertial measurement unit 63.

[0032] Next, in step S22, the control device 12 acquires map image data from the camera 41. The image data acquired from the camera 41 and used in the map data M1 is the map image data. Steps S21 and S22 are performed synchronously. Therefore, the location information acquired in step S21 includes the coordinates in the map coordinate system of the point where the map image data was acquired.

[0033] Next, in step S23, the control device 12 links the location information acquired in step S21 to the map image data acquired in step S22. The control device 12 repeatedly performs the processes of steps S21 to S23. As a result, the control device 12 acquires map image data from the camera 41 multiple times. Each piece of map image data is associated with location information. When the vehicle 10 has completed traveling within the range for which the map data M1 is to be created, the map creation process ends. The map data M1 created by the map creation process is data before a reliability evaluation value is assigned. The map data M1 before a reliability evaluation value is assigned is referred to as pre-evaluation map data.

[0034] As shown in Fig. 4, when the control device 12 finishes the map creation process, it performs a reliability evaluation process in step S12. In the reliability evaluation process, the reliability of the map image data is evaluated. The reliability is a reliability related to the accuracy of self-location estimation. The reliability evaluation process will be described in detail.

[0035] As shown in FIG. 6, in step S31, the control device 12 determines map image data to be evaluated. The map image data to be evaluated is map image data that is included in the pre-evaluation map data created in the map creation process and has not yet been evaluated for reliability. In steps S32 to S36, the reliability of the map image data is evaluated. The map image data evaluated in steps S32 to S36 is the map image data to be evaluated. By sequentially changing the map image data to be evaluated in step S31, the reliability of all map image data included in the pre-evaluation map data is evaluated.

[0036] In step S32, the control device 12 performs pixel value evaluation. In pixel value evaluation, the reliability of the map image data is evaluated based on the luminance of the pixels that make up the map image data. The control device 12 determines whether the luminance of the pixels that make up the map image data is less than a low-luminance threshold. An excessively low luminance of the pixels that make up the map image data can make it difficult to match the image data with the map image data when estimating the self-location, or can cause erroneous matching. The low-luminance threshold is set based on the luminance that is acceptable when matching the image data with the map image data. Examples of cases in which the luminance of the pixels that make up the map image data falls below the low-luminance threshold include when the road surface Sr is dirty, when there is a shadow on the road surface Sr, and when there is insufficient illuminance. Insufficient illuminance occurs, for example, when the road surface Sr is wet during rainy weather, when the lighting device 51 breaks down, or when the road surface Sr has a color that does not reflect light well.

[0037] The control device 12 determines whether the brightness of pixels constituting the map image data is equal to or greater than a high brightness threshold. Excessively high brightness of pixels constituting the map image data may make it difficult to match image data with map image data when estimating the vehicle's position, or may result in erroneous matching. The high brightness threshold is set based on the brightness that is acceptable when matching image data with map image data. Examples of cases in which the brightness of pixels constituting the map image data is equal to or greater than the high brightness threshold include when sunlight is shining below the vehicle 10, or when the road surface Sr is painted to make it more reflective of light.

[0038] The control device 12 reduces the reliability of the map image data when the total number of pixels constituting the map image data, including those with brightness below the low brightness threshold and those with brightness equal to or greater than the high brightness threshold, is equal to or greater than a pixel threshold. Specifically, if the reliability evaluation value at the time of performing step S32 corresponds to normal-quality image data, the control device 12 reduces the evaluation value corresponding to normal-quality image data to that corresponding to low-quality image data. If the reliability evaluation value at the time of performing step S32 corresponds to low-quality image data, the control device 12 maintains the reliability evaluation value. As described above, pixels with brightness below the low brightness threshold and pixels with brightness equal to or greater than the high brightness threshold can make matching difficult or can cause erroneous matching. The pixel threshold is set based on the number of pixels allowed when matching image data with map image data.

[0039] Next, in step S33, the control device 12 performs feature point number evaluation. In feature point number evaluation, the reliability of the map image data is evaluated from the number of feature points included in the map image data. The control device 12 extracts feature points from the map image data. The feature points can be extracted using the same method as when performing self-location estimation. The feature points are extracted based on a brightness gradient. If the number of feature points included in the map image data is less than the feature point threshold, the control device 12 reduces the reliability of the map image data. If the number of feature points included in the map image data is small, there will be fewer feature point pairs when matching the image data with the map image data. In this case, matching will be difficult. The feature point threshold is set so that the number of feature point pairs is within an acceptable range when matching with the image data.

[0040] Next, in step S34, the control device 12 performs a feature point distribution evaluation. In the feature point distribution evaluation, the reliability of the map image data is evaluated based on the distribution of feature points included in the map image data. The control device 12 determines whether the distribution of feature points is biased. The bias in the distribution of feature points can be calculated, for example, by the standard deviation. The bias in the distribution of feature points includes the feature points being distributed in a specific shape, such as a linear distribution. The bias in the distribution of feature points includes the feature points being concentrated in a part of the map image data. Examples of cases in which the distribution of feature points is biased include when a grating is captured, when a boundary line between painted and unpainted portions of the road surface Sr is captured, and when a shadow and a sunny spot are captured. The bias in the distribution of feature points can make it difficult to match image data with map image data or can cause erroneous matching. If the feature points included in the map image data are distributed biased, the control device 12 reduces the reliability of the map image data.

[0041] Next, in step S35, the control device 12 performs evaluation by matching map image data with each other. More specifically, the control device 12 matches the map image data to be evaluated with all map image data different from the map image data to be evaluated. The matching is similar to the matching performed during self-location estimation. That is, the control device 12 searches for pairs of feature points between the map image data. If the number of matching feature points is equal to or greater than a first matching threshold, the control device 12 reduces the reliability of the map image data. The number of matching feature points is the number of pairs of feature points. The first matching threshold is set so that when matching with image data is performed, multiple map image data are not identified as map image data corresponding to the image data.

[0042] Next, in step S36, the control device 12 matches the sample image data with the map image data. The sample image data is image data whose reliability evaluation value is less than a threshold. In other words, the sample image data is a sample of low-quality image data. Examples of sample image data include image data that shows gratings, image data that shows painted surfaces such as road markings, image data that shows shadows and sunny areas, image data that shows dirt, image data that shows areas where specular reflection occurs such as puddles, and image data with insufficient illumination. The control device 12 matches the sample image data with the map image data. The matching is similar to the matching performed during self-location estimation. In other words, the control device 12 searches for pairs of feature points in the map image data and the sample image data. If the number of matching feature points is equal to or greater than a second matching threshold, the control device 12 reduces the reliability of the map image data. If there are multiple sample image data, the control device 12 matches each of the multiple sample image data with the map image data. In this case, the control device 12 reduces the reliability of the map image data if the number of matching feature points between at least one sample image data and the map image data is equal to or greater than a second matching threshold. The second matching threshold may be the same as or different from the first matching threshold. The second matching threshold is set so that when matching is performed between low-quality image data and map image data, the map image data is identified as map image data corresponding to the low-quality image data. After completing the processing of step S36, the control device 12 ends the reliability evaluation processing.

[0043] As shown in Fig. 4, when the control device 12 completes the reliability evaluation process, it assigns a reliability evaluation value to the map image data in step S13. In this embodiment, an evaluation value corresponding to low-quality image data is assigned to map image data whose reliability has been reduced in at least one of steps S32 to S36. An evaluation value corresponding to normal-quality image data is assigned to map image data whose reliability has not been reduced in any of steps S32 to S36. When the control device 12 completes the process in step S13, it terminates map data creation control.

[0044] As described above, the control device 12 that creates the map data M1 by performing map data creation control is the map data creation device. [Effects of this embodiment] (1) When creating the map data M1, the control device 12 evaluates the reliability of the map image data. The control device 12 then assigns a reliability evaluation value to the map image data. When estimating the self-location, the control device 12 excludes map image data whose reliability evaluation value is less than a threshold. In this way, the reliability can be used as an indicator for determining whether to exclude map image data when estimating the self-location. By estimating the self-location while excluding map image data whose reliability evaluation value is less than the threshold, it is possible to prevent matching from failing or erroneous matching from occurring due to map image data whose reliability evaluation value is less than the threshold. Therefore, the accuracy of the self-location estimation can be improved.

[0045] (2) The control device 12 reduces the reliability of map image data when the total number of pixels constituting the map image data, including those with brightness levels below the low brightness threshold and those with brightness levels above the high brightness threshold, is equal to or exceeds the pixel threshold. Map image data in which the total number of pixels with brightness levels below the low brightness threshold and those with brightness levels above the high brightness threshold is equal to or exceeds the pixel threshold, can make matching difficult or can cause erroneous matching. By reducing the reliability of such map image data, matching failures or erroneous matching caused by map image data with a reliability evaluation value below the threshold are prevented.

[0046] (3) When the number of feature points included in the map image data is less than the feature point threshold, the control device 12 reduces the reliability of the map image data. When the number of feature points included in the map image data is less than the feature point threshold, the number of feature point pairs decreases when matching is performed between the image data and the map image data. By reducing the reliability of such map image data, it is possible to prevent matching from becoming impossible due to map image data with a reliability evaluation value less than the threshold.

[0047] (4) The control device 12 reduces the reliability of map image data when the feature points contained in the map image data are unevenly distributed. Map image data that shows gratings, map image data that shows the boundary between painted and unpainted portions of the road surface Sr, and map image data that shows shadows and sunny areas can make matching difficult or cause erroneous matching. By reducing the reliability of such map image data, matching failures or erroneous matching caused by map image data with a reliability evaluation value below a threshold are prevented.

[0048] (5) The control device 12 performs matching between multiple map image data included in the map data M1, and when the number of matching feature points is equal to or greater than the first matching threshold, reduces the reliability of the map image data. If the number of matching feature points is large when matching between map image data, multiple map image data may be identified corresponding to the image data during self-location estimation. In this case, it may be difficult for the control device 12 to determine which of the multiple map image data is correct, which may reduce the accuracy of self-location estimation. By reducing the reliability of such map image data, the accuracy of self-location estimation can be improved.

[0049] (6) The control device 12 matches the sample image data with the map image data, and if the number of matching feature points is equal to or greater than the second matching threshold, reduces the reliability of the map image data. If the number of matching feature points is equal to or greater than the second matching threshold when matching the sample image data with the map image data, the map image data is map image data whose features match those of the low-quality image data. By regarding such map image data as low-quality image data and reducing its reliability, matching failures or erroneous matching caused by map image data with a reliability evaluation value below the threshold are prevented.

[0050] [Example of change] The embodiment can be modified as follows: The embodiment and the following modifications can be combined with each other to the extent that they are not technically inconsistent.

[0051] The control device 12 may reacquire map image data whose reliability evaluation value is less than a threshold value in the map data creation control. When the vehicle 10 moves to coordinates in the map coordinate system associated with map image data whose reliability evaluation value is less than a threshold value, the control device 12 acquires map image data from the camera 41. The vehicle 10 may be driven by a human or controlled by the control device 12. In this way, the reliability can be used as an indicator of whether or not to reacquire map image data.

[0052] The control device 12 may use the reliability evaluation value assigned to the map image data when determining the reliability of the self-location estimation. In this case, the higher the reliability evaluation value assigned to the map image data, the higher the reliability of the self-location estimation.

[0053] The control device 12 may identify a low-quality area from the map image data. The low-quality area is a part of the map image data that may cause matching to be impossible or result in erroneous matching when matching image data with the map image data. The control device 12 may, for example, divide the map image data into multiple areas and evaluate the reliability of each area. The control device 12 then defines the area where the reliability evaluation value is less than a threshold as the low-quality area. When performing self-location estimation, the control device 12 masks the low-quality area and performs matching with the image data. Feature points are not extracted from the masked area. This makes it possible to prevent the accuracy of self-location estimation from being reduced due to the low-quality area.

[0054] The control device 12 may perform at least one of pixel value evaluation, feature point number evaluation, feature point distribution evaluation, evaluation by matching between map image data, and evaluation by matching with sample image data. The control device 12 may perform reliability evaluation processing using any combination of pixel value evaluation, feature point number evaluation, feature point distribution evaluation, evaluation by matching between map image data, and evaluation by matching with sample image data.

[0055] The control device 12 may assign an evaluation value corresponding to low-quality image data to map image data whose reliability has been reduced by multiple evaluations among pixel value evaluation, feature point count evaluation, feature point distribution evaluation, evaluation by matching between map image data, and evaluation by matching with sample image data. The control device 12 may assign an evaluation value corresponding to low-quality image data to map image data whose reliability has been reduced by two, three, four, or five evaluations.

[0056] If the reliability evaluation value is expressed as a score, the control device 12 may lower the score each time the reliability decreases in each of the pixel value evaluation, feature point number evaluation, feature point distribution evaluation, evaluation by matching between map image data, and evaluation by matching with sample image data.

[0057] The map data creation device may include the control device 12 and an external device different from the control device 12. The external device is a device provided outside the vehicle 10. The hardware configuration of the external device is, for example, the same as that of the control device 12. In this case, the control device 12 creates pre-evaluation map data. The pre-evaluation map data created by the control device 12 is stored in the external device through communication between the control device 12 and the external device or via a storage medium. The external device evaluates the pre-evaluation map data through a reliability evaluation process. As a result, the external device creates map data M1.

[0058] When the reliability evaluation value is a score, a threshold value may be set for the score, which is set so as to exclude low-quality image data. The map data M1 may be created by driving a vehicle other than the vehicle 10. That is, the vehicle used to create the map data M1 may be different from the autonomous vehicle for which the map data M1 is used. The vehicle used to create the map data M1 may be a vehicle that cannot drive autonomously.

[0059] The imaging device may be a linear image sensor. The storage device in which the map data M1 is stored may be the storage unit 14. The technical ideas that can be understood from the above-described embodiment and modified examples will be described.

[0060] [Aspect 1] A map data creation device that creates map data used by an autonomous vehicle when estimating its own position, and that is used to compare with image data acquired from an imaging device equipped in the autonomous vehicle, the map data creation device having a control device that acquires map image data obtained by capturing images of the road surface using an imaging device equipped in the vehicle, links location information including the coordinates of the point where the map image data was acquired to the map image data, and evaluates the reliability of the map image data, which is the reliability related to the accuracy of the self-position estimation.

[0061] [Aspect 2] The map data creation device described in [Aspect 1], wherein the control device reduces the reliability of the map image data when the total number of pixels constituting the map image data, including those whose brightness is less than a low brightness threshold and those whose brightness is equal to or greater than a high brightness threshold, is equal to or greater than a pixel threshold.

[0062] [Aspect 3] The map data creation device described in [Aspect 1] or [Aspect 2], wherein the control device reduces the reliability of the map image data when the number of feature points contained in the map image data is less than a feature point threshold.

[0063] [Aspect 4] A map data creation device described in any one of [Aspect 1] to [Aspect 3], wherein the control device reduces the reliability of the map image data when feature points contained in the map image data are unevenly distributed.

[0064] [Aspect 5] A map data creation device described in any one of [Aspect 1] to [Aspect 4], wherein the control device matches multiple pieces of map image data with each other, and if the number of matching feature points is greater than or equal to a first matching threshold, reduces the reliability of the map image data.

[0065] [Aspect 6] The map data creation device according to any one of [Aspects 1] to [Aspect 5], wherein the control device matches the map image data with sample image data for which the reliability evaluation value is less than a threshold, and reduces the reliability of the map image data when the number of matching feature points is equal to or greater than a second matching threshold. [Explanation of symbols]

[0066] M1...map data, Sr...road surface, 10...vehicle which is an autonomous vehicle, 12...control device which is a map data creation device, 41...camera which is an imaging device.

Claims

1. 1. A map data creation device that creates map data used by an autonomous vehicle when estimating its own position, the map data being used for matching with image data acquired from an imaging device provided in the autonomous vehicle, The map data creation device includes a control device, The control device acquiring map image data obtained by capturing an image of a road surface using an imaging device provided in the vehicle; Linking location information including coordinates of the point where the map image data was acquired to the map image data; Evaluating the reliability of the map image data, which is a reliability related to the accuracy of the self-location estimation; The map data creation device performs matching between the plurality of map image data, and reduces the reliability of the map image data when the number of matching feature points is equal to or greater than a first matching threshold.

2. 1. A map data creation device that creates map data used by an autonomous vehicle when estimating its own position, the map data being used for matching with image data acquired from an imaging device provided in the autonomous vehicle, The map data creation device includes a control device, The control device acquiring map image data obtained by capturing an image of a road surface using an imaging device provided in the vehicle; Linking location information including coordinates of the point where the map image data was acquired to the map image data; Evaluating the reliability of the map image data, which is a reliability related to the accuracy of the self-location estimation; A map data creation device that matches sample image data whose reliability evaluation value is less than a threshold with the map image data, and if the number of matching feature points is greater than or equal to a second matching threshold, reduces the reliability of the map image data.

Images