Determination device

Abstract

To detect that an actual position of a feature does not match a position in map data and corrects the map data according to the needs.SOLUTION: A determination device acquires first distance information indicating distance from a mobile body to a target, measured by a measuring unit, and second distance information indicating distance from the mobile body to the target, estimated by an estimation unit. The determination device also acquires self-position accuracy information of the mobile body. The determination device then determines whether there is a deviation equal to or more than a predetermined value between a position of the target in map information and a position of the target in a real environment based on a difference value between positions indicated by the first distance information and the second distance information and the self-position accuracy information.SELECTED DRAWING: Figure 7

Description

[Technical Field]

[0001] The present invention relates to a technique for correcting the position of a feature in map data. [Background technology]

[0002] A method has been proposed for updating map data based on surrounding information detected by external sensors mounted on a vehicle. For example, Patent Document 1 describes a method for generating change point candidate data by comparing surrounding information based on the output of a sensor unit with a partial map DB, and updating the map data while taking into account the accuracy of the sensor when the change point candidate data was acquired. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2016-156973 Summary of the Invention [Problem to be solved by the invention]

[0004] The map data stored in the server includes the position coordinates of features. However, if the position coordinates of the features change for some reason after they are measured, the position coordinates of the features included in the map data will no longer match the actual position coordinates of the features. For example, if a sign's position shifts or an installed object on the ground moves due to a vehicle collision, the actual position of the feature will no longer match the position in the map data. In such a situation, if the position coordinates of the features in the map data are used to estimate the vehicle's own position, the estimation error will be large, which may interfere with driving assistance or autonomous driving.

[0005] The above is an example of a problem that the present invention aims to solve. The present invention aims to detect when the actual position of a feature does not match the position in map data, and to correct the map data as necessary. [Means for solving the problem]

[0006] The invention described in the claims is a determination device comprising: a first acquisition unit that acquires first distance information indicating the distance from a moving body to an object measured by a measurement unit, and second distance information indicating the distance from the position of the moving body to the position of the object in map information estimated by an estimation unit; a second acquisition unit that acquires measurement accuracy information indicating the measurement accuracy of the position of the object by the measurement unit, and estimation accuracy information indicating the estimation accuracy of the position of the moving body by the estimation unit; and a determination unit that determines whether or not there is a deviation of a predetermined value or more between the position of the object in the map information and the position of the object in the real environment based on the difference value of the distance indicated by the first distance information and the second distance information, the measurement accuracy information, and the estimation accuracy information.

[0007] The invention described in the claims is a determination method executed by a determination device, characterized in that it comprises: a first acquisition process for acquiring first distance information indicating the distance from a moving body to an object, measured by a measurement unit, and second distance information indicating the distance from the position of the moving body to the position of the object in map information, estimated by an estimation unit; a second acquisition process for acquiring measurement accuracy information indicating the measurement accuracy of the position of the object by the measurement unit, and estimation accuracy information indicating the estimation accuracy of the position of the moving body by the estimation unit; and a determination process for determining whether or not there is a deviation of a predetermined value or more between the position of the object in the map information and the position of the object in the real environment, based on the difference value of the distance indicated by the first distance information and the second distance information, the measurement accuracy information, and the estimation accuracy information.

[0008] The invention described in the claims is a program executed by a determination device having a computer, characterized in that the computer functions as a first acquisition unit that acquires first distance information indicating the distance from a moving body to an object measured by a measurement unit and second distance information indicating the distance from the position of the moving body to the position of the object in map information estimated by an estimation unit; a second acquisition unit that acquires measurement accuracy information indicating the measurement accuracy of the position of the object by the measurement unit and estimation accuracy information indicating the estimation accuracy of the position of the moving body by the estimation unit; and a determination unit that determines whether or not there is a deviation of a predetermined value or more between the position of the object in the map information and the position of the object in the real environment based on the difference value of the distance indicated by the first distance information and the second distance information, the measurement accuracy information, and the estimation accuracy information.

[0009] The invention described in the claims is a determination device comprising: a first acquisition unit that acquires first distance information indicating the distance from a moving body to an object measured by a measurement unit, and second distance information indicating the distance from the position of the moving body to the position of the object in map information estimated by an estimation unit; a second acquisition unit that acquires measurement accuracy information indicating the measurement accuracy of the position of the object by the measurement unit, and estimation accuracy information indicating the estimation accuracy of the position of the moving body by the estimation unit; and a determination unit that determines whether or not there is a deviation of a predetermined value or more between the position of the object in the map information and the position of the object based on the first distance, based on the difference value of the distance indicated by the first distance information and the second distance information, the measurement accuracy information, and the estimation accuracy information. [Brief explanation of the drawings]

[0010] [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] FIG. 2 is a block diagram showing a functional configuration of a server device. [Figure 4] This is a diagram showing a state variable vector in two-dimensional orthogonal coordinates. [Figure 5]FIG. 10 is a diagram illustrating a schematic relationship between a prediction step and a measurement update step. [Figure 6] This shows the positional relationship between the vehicle and features at the time of position estimation. [Figure 7] 10 is a flowchart of a difference value evaluation process according to the first embodiment. [Figure 8] 4 is a flowchart of a map correction process according to the first embodiment. [Figure 9] This is a diagram showing a state variable vector in three-dimensional orthogonal coordinates. [Figure 10] FIG. 10 is a diagram illustrating a schematic relationship between a prediction step and a measurement update step. [Figure 11] This shows the positional relationship between the vehicle and features at the time of position estimation. [Figure 12] 10 is a flowchart of a difference value evaluation process according to a second embodiment. [Figure 13] 10 is a flowchart of a map correction process according to a second embodiment. [Figure 14] FIG. 10 is a diagram illustrating the effects of the embodiment. [Figure 15] An example of multiple road signs installed side by side is shown below. [Figure 16] Examples of composite white lines and white lines with road studs are shown below. DETAILED DESCRIPTION OF THE INVENTION

[0011] In one preferred embodiment of the present invention, the determination device includes a first acquisition unit that acquires first distance information indicating the distance from a moving body to an object, measured by a measurement unit, and second distance information indicating the distance from the moving body to the object, estimated by an estimation unit; a second acquisition unit that acquires self-position accuracy information of the moving body; and a determination unit that determines whether or not there is a deviation of a predetermined value or more between the position of the object in the map information and the position of the object in the real environment, based on the difference value between the distances indicated by the first distance information and the second distance information, and the self-position accuracy information.

[0012] The above-mentioned determination device acquires first distance information, measured by a measurement unit, indicating the distance from the moving body to the object, and second distance information, estimated by an estimation unit, indicating the distance from the moving body to the object. The determination device also acquires self-location accuracy information of the moving body. Then, based on a difference value between the positions indicated by the first distance information and the second distance information and the self-location accuracy information, it determines whether or not there is a deviation of a predetermined value or more between the position of the object in the map information and the position of the object in the real environment. The above-mentioned determination device determines whether or not there is a deviation between the position of the object in the map information and the actual position of the object measured by the measurement unit.

[0013] In one aspect of the above determination device, the second acquisition unit further acquires measurement accuracy information of the measurement unit, and the determination unit determines whether or not there is a deviation of the target position equal to or greater than the predetermined value based on the self-location accuracy information and the measurement accuracy information. In this aspect, the position deviation of the target object is determined taking into account the measurement accuracy of the measurement unit.

[0014] Another aspect of the above determination device includes a third acquisition unit that acquires error information including the difference value, the self-location accuracy information, and measurement accuracy information for an object determined to have a deviation of equal to or greater than the predetermined value, and a correction unit that, when the third acquisition unit acquires a predetermined number of pieces of error information for the same object, corrects the position of the object in the map information by statistically processing the difference values ​​included in the acquired error information. In this aspect, the position of the object in the map information can be corrected by statistically processing the difference values ​​included in the error information.

[0015] In another aspect of the above determination device, the correction unit converts the difference value from a moving body coordinate system to a map coordinate system, and corrects the position in the map information.

[0016] In another aspect of the above determination device, when the third acquisition unit acquires error information for the same object from a predetermined number of vehicles or more, and when a variation in the difference values ​​included in the acquired error information is a predetermined value or more, the correction unit lowers the level of the position accuracy information assigned to the object in the map information. In this aspect, information indicating low position accuracy is assigned to the object in the map information.

[0017] In another aspect of the above-described determination device, the determination device is mounted on a vehicle and includes a transmitter that, when the determination unit determines that there is a deviation of the predetermined value or more, transmits error information including the difference value, the self-position accuracy information, and the measurement accuracy information to an external server. In this aspect, when it is determined that there is a deviation of the position of the object that is a predetermined value or more, the error information is transmitted to the server.

[0018] In another aspect of the above-mentioned determination device, when a plurality of pieces of error information are acquired for the same object, and a variance value based on the plurality of pieces of error information is equal to or less than a predetermined value, the transmitter transmits the plurality of pieces of error information or information obtained using the plurality of pieces of error information to the external server. In this aspect, the error information etc. is transmitted only when there is a high possibility that the position of the feature is misaligned, thereby reducing the frequency of transmission.

[0019] In another aspect of the above determination device, the error information includes date and time information and identification information of the object. In this aspect, it is possible to know when the positional deviation of the object occurred based on the date and time information.

[0020] In another aspect of the above determination device, the measurement unit and the estimation unit are mounted on a vehicle, the determination device is installed in a server, and the first acquisition unit acquires the first distance information and the second distance information from the vehicle. In this aspect, the determination device installed in the server can determine the positional deviation of the object based on the information acquired from the vehicle.

[0021] In another preferred embodiment of the present invention, a determination method executed by a determination device includes a first acquisition step of acquiring first distance information measured by a measurement unit and indicating a distance from a moving body to an object and second distance information estimated by an estimation unit and indicating a distance from the moving body to the object, a second acquisition step of acquiring self-location accuracy information of the moving body, and a determination step of determining whether or not there is a deviation of a predetermined value or more between the position of the object in map information and the position of the object in a real environment, based on a difference value between the distances indicated by the first distance information and the second distance information and the self-location accuracy information. According to this method, whether or not there is a deviation between the position of the object in the map information and the actual position of the object measured by a measurement unit is determined.

[0022] In another preferred embodiment of the present invention, a program executed by a determination device including a computer causes the computer to function as a first acquisition unit that acquires first distance information indicating the distance from a moving object, measured by a measurement unit, and second distance information indicating the distance from the moving object to the object, estimated by an estimation unit; a second acquisition unit that acquires self-location accuracy information of the moving object; and a determination unit that determines whether or not there is a deviation of a predetermined value or more between the position of the object in map information and the position of the object in the real environment, based on a difference value between the distances indicated by the first distance information and the second distance information, and the self-location accuracy information. The above-mentioned determination device can be realized by executing this program on a computer. This program can be stored and handled in a storage medium.

[0023] In another preferred embodiment of the present invention, the determination device includes a first acquisition unit that acquires first distance information indicating the distance from a moving body to an object measured by a measurement unit and second distance information indicating the distance from the moving body to the object estimated by an estimation unit, a second acquisition unit that acquires self-position accuracy information of the moving body, and a determination unit that determines whether or not there is a deviation of a predetermined value or more between the position of the object in map information and the position of the object based on the first distance, based on the difference value of the distance indicated by the first distance information and the second distance information, and the self-position accuracy information.

[0024] The above-mentioned determination device acquires first distance information, measured by a measurement unit, indicating the distance from the moving body to the object, and second distance information, estimated by an estimation unit, indicating the distance from the moving body to the object. The determination device also acquires self-location accuracy information of the moving body. Then, based on a difference value between the positions indicated by the first distance information and the second distance information and the self-location accuracy information, it determines whether or not there is a deviation of a predetermined value or more between the position of the object in the map information and the position of the object based on the first distance information. The above-mentioned determination device determines whether or not there is a deviation between the position of the object in the map information and the actual position of the object measured by the measurement unit. [Example]

[0025] Preferred embodiments of the present invention will now be described with reference to the drawings. [Schematic configuration]

[0026] FIG. 1 is a schematic diagram of a driving assistance system according to this embodiment. The driving assistance system is broadly divided into an on-board device 10 mounted on a vehicle 1 and a server device 20. The vehicle 1 is equipped with the on-board device 10 that controls driving assistance for the vehicle, 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. The server device 20 includes a map database (hereinafter, "database" will be referred to as "DB") 23 that stores map data. The on-board device 10 and the server device 20 transmit and receive data via wireless communication. Note that for convenience, only one vehicle 1 and on-board device 10 are shown in FIG. 1, but in reality, the server device 20 communicates with the on-board devices 10 of multiple vehicles 1.

[0027] The vehicle-mounted device 10 is electrically connected to the LIDAR 2, the gyro sensor 3, the vehicle speed sensor 4, and the GPS receiver 5, and estimates the position of the vehicle on which the vehicle-mounted device 10 is mounted (also referred to as the "vehicle position") based on the outputs of these sensors. Then, based on the result of estimating the vehicle position, the vehicle-mounted device 10 performs automatic driving control of the vehicle so that the vehicle travels along a route to a set destination. The vehicle-mounted device 10 stores a map DB 13 that stores road data and feature information, which is information about landmark features installed near the road. The above-mentioned landmark features are, for example, kilometer posts, 100-meter posts, delineators, traffic infrastructure facilities (e.g., signs, directional signs, traffic lights), utility poles, streetlights, and other features that are periodically lined along the side of the road. The feature information is information in which at least a feature ID, which is identification information for each feature, location information of the feature, and orientation information of the feature are associated with each other. Then, the vehicle-mounted device 10 estimates the vehicle position based on this feature information by comparing it with the output of the LIDAR 2 and the like.

[0028] The LIDAR 2 emits a pulsed laser beam over a predetermined angular range in the horizontal and vertical directions to discretely measure the distance to an object in the external environment and generate three-dimensional point cloud information indicating the position of the object. In this example, the LIDAR 2 includes an irradiation unit that irradiates a laser beam while changing the irradiation direction, a light receiving unit that receives reflected (scattered) light from the irradiated laser beam, and an output unit that outputs scan data based on a light receiving signal output by the light receiving unit. The scan data is 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. In this example, the LIDAR 2 is installed facing the vehicle's traveling direction so as to scan at least the area ahead of the vehicle. The LIDAR 2, gyro sensor 3, vehicle speed sensor 4, and GPS receiver 5 each provide output data to the vehicle-mounted device 10.

[0029] 2 is a block diagram showing the functional configuration of the vehicle-mounted device 10. The vehicle-mounted device 10 mainly includes an interface 11, a storage unit 12, an input unit 14, a control unit 15, and an information output unit 16. These elements are interconnected via a bus line.

[0030] 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 control unit 15. The interface 11 also supplies signals related to vehicle driving control generated by the control unit 15 to an electronic control unit (ECU) of the vehicle.

[0031] The storage unit 12 stores programs executed by the control unit 15 and information required for the control unit 15 to execute predetermined processes. In this embodiment, the storage unit 12 also stores a map DB 13 including feature information. The feature information is information in which information related to each feature is associated with the feature, and here includes a feature ID, which is identification information for the feature, location information, and shape information. The location information indicates the absolute position of the feature, expressed by latitude and longitude (and altitude), etc. The shape information is information related to the shape of the feature, and includes normal information indicating the orientation of the feature (i.e., the normal direction relative to the front) and size information indicating the size of the feature. Note that the map DB 13 may be updated periodically. In this case, for example, the control unit 15 receives partial map information related to the area to which the vehicle position belongs from a server device 20 that manages map information via a communication unit (not shown), and reflects the partial map information in the map DB 13.

[0032] The input unit 14 is a button, touch panel, remote controller, voice input device, etc. for user operation, and accepts inputs to specify a destination for route search, inputs to specify whether autonomous driving is on or off, etc. The information output unit 16 is, for example, a display, speaker, etc. that outputs information based on the control of the control unit 15.

[0033] The control unit 15 includes a CPU that executes programs and controls the entire in-vehicle device 10. In this embodiment, the control unit 15 includes a vehicle position estimation unit 17 and an autonomous driving control unit 18. The vehicle position estimation unit 17 corrects the vehicle position estimated from the output data of the gyro sensor 3, the vehicle speed sensor 4, and / or the GPS receiver 5, based on the distance and angle measurement values ​​relative to features by the LIDAR 2 and the position information of the features extracted from the map DB 13. In this embodiment, as an example, the vehicle position estimation unit 17 alternately executes a prediction step of estimating the vehicle position from the output data of the gyro sensor 3, the vehicle speed sensor 4, etc., based on a state estimation method based on Bayesian estimation, and a measurement update step of correcting the estimated value of the vehicle position calculated in the immediately preceding prediction step.

[0034] The autonomous driving control unit 18 refers to the map DB 13 and transmits signals necessary for autonomous driving control to the vehicle based on the set route and the vehicle position estimated by the vehicle position estimation unit 17. The autonomous driving control unit 18 sets a target trajectory based on the set route, and controls the position of the vehicle by transmitting a guide signal to the vehicle so that the vehicle position estimated by the vehicle position estimation unit 17 deviates from the target trajectory by a predetermined amount or less.

[0035] In the above configuration, the lidar 2 is an example of a measurement unit of the present invention, and the control unit 15 is an example of an estimation unit, a first acquisition unit, a second acquisition unit, and a determination unit of the present invention.

[0036] FIG. 3 is a block diagram showing the functional configuration of the vehicle-mounted device server device 20. The server device 20 mainly includes a communication unit 21, a storage unit 22, and a control unit 25. These elements are interconnected via a bus line. The communication unit 21 communicates with the vehicle-mounted device 10 via wireless communication or the like. The storage unit 22 stores programs executed by the control unit 25 and information required for the control unit 25 to execute predetermined processes. The storage unit 22 also stores a map DB 23 including feature information. The control unit 25 includes a CPU that executes programs and controls the entire server device 20. Specifically, the control unit 25 executes a map correction process, which will be described later.

[0037] [First Example] (Vehicle position estimation process) Next, a first embodiment of the process of estimating the vehicle position by the vehicle position estimation unit 17 will be described. The vehicle position estimation unit 17 performs vehicle position estimation by sequentially repeating a prediction step and a measurement update step. As the state estimation filter used in these steps, various filters developed to perform Bayesian estimation can be used, such as an extended Kalman filter, an unscented Kalman filter, and a particle filter. As such, various methods have been proposed for position estimation based on Bayesian estimation. Below, a brief description will be given of vehicle position estimation using an extended Kalman filter as an example.

[0038] FIG. 4 is a diagram showing the state variable vector X in two-dimensional Cartesian coordinates. In the first embodiment, the z-coordinate is assumed to be projected onto the two-dimensional Cartesian coordinates of x and y. As shown in FIG. 4, the vehicle position on a plane defined on the two-dimensional Cartesian coordinates of x and y is represented by the coordinates "(x, y)" and the vehicle's orientation "Ψ." Here, the orientation Ψ is defined as the angle between the vehicle's traveling direction and the x-axis. The coordinates (x, y) indicate an absolute position equivalent to, for example, a combination of latitude and longitude.

[0039] FIG. 5 is a diagram showing the general relationship between the prediction step and the measurement update step. As shown in FIG. 5, the prediction step and the measurement update step are repeated to sequentially calculate and update the estimated value of the state variable vector X. Here, the state variable vector at the reference time (i.e., the current time) "k" to be calculated is expressed as "Xe(k)" or "Xp(k)". The provisional predicted value predicted in the prediction step is given the subscript "e", and the more accurate estimated value updated in the measurement update step is given the subscript "p". ("State variable vector Xe(k)=(x e (k), y e (k), Ψ e (k)) T ", "State variable vector Xp(k)=(x p (k), y p (k), Ψ p (k)) T" is written as ".)

[0040] In the prediction step, the vehicle position estimation unit 17 calculates a predicted value of the vehicle position at time k (also referred to as the "predicted vehicle position") Xe(k) by applying the vehicle's moving speed "v" and the yaw angular velocity "ω" about the z-axis (collectively referred to as the "control value u(k)") to the state variable vector Xp(k-1) at time k-1 calculated in the immediately preceding measurement update step. At the same time, the vehicle position estimation unit 17 calculates a covariance matrix Pe(k) corresponding to the error distribution of the predicted vehicle position Xe(k) from the covariance matrix Pp(k-1) at time k-1 calculated in the immediately preceding measurement update step.

[0041] In the measurement update step, the vehicle position estimation unit 17 associates the position vectors of the features registered in the map DB 13 with the scan data of the LIDAR 2. When the vehicle position estimation unit 17 has established this association, it acquires the measurement value (referred to as the "feature measurement value") "Z(i)" of the associated feature by the LIDAR 2 and the predicted value (referred to as the "feature predicted value") "Ze(i)" of the feature obtained by modeling the measurement process by the LIDAR 2 using the predicted vehicle position Xe(k) and the position vector of the feature registered in the map DB 13. The feature measurement value Z(i) is a two-dimensional vector obtained by converting the distance and scan angle of the feature measured by the LIDAR 2 at time i into components whose axes are the vehicle's traveling direction and lateral direction. The vehicle position estimation unit 17 then calculates the difference between the feature measurement value Z(i) and the feature predicted value Ze(i) as shown in the following equation (1).

number

[0042] Furthermore, the vehicle position estimation unit 17 multiplies the difference between the feature measurement value Z(i) and the feature prediction value Ze(i) by the Kalman gain "K(k)" and adds this to the predicted vehicle position Xe(k), as shown in the following equation (2), to calculate an updated state variable vector (also called "measurement updated vehicle position") Xp(k).

number

[0043] In the measurement update step, the vehicle position estimation unit 17 calculates a covariance matrix Pp(k) corresponding to the error distribution of the measurement update vehicle position Xp(k) from the prior covariance matrix Pe(k), similar to the prediction step. Parameters such as the Kalman gain K(k) can be calculated in the same manner as in a known vehicle positioning technique using an extended Kalman filter, for example.

[0044] When the vehicle position estimation unit 17 is able to associate the position vectors of multiple features registered in the map DB 13 with the scan data of the LIDAR 2, the vehicle position estimation unit 17 may perform the measurement update step based on any one selected feature measurement value, or may perform the measurement update step multiple times based on all of the associated feature measurement values. When using multiple feature measurement values, the vehicle position estimation unit 17 takes into account that the LIDAR measurement accuracy deteriorates the further a feature is from the LIDAR 2, and decreases the weighting for that feature the longer the distance between the LIDAR 2 and the feature. Generally, the major axis, minor axis, and angle of an error ellipse indicating the expected error range (for example, a range based on a 99% confidence interval) can be calculated from the eigenvalues ​​and eigenvectors of a covariance matrix corresponding to the above-mentioned error distribution.

[0045] In this way, the prediction step and the measurement update step are repeatedly performed, and the predicted vehicle position Xe(k) and the measurement update vehicle position Xp(k) are calculated sequentially, thereby calculating the most probable vehicle position. The vehicle position calculated in this way is called the "estimated vehicle position."

[0046] In the above explanation, the feature measurement value is an example of first distance information of the present invention, the feature prediction value is an example of second distance information of the present invention, the position estimation accuracy is an example of self-position accuracy information of the present invention, and the lidar measurement accuracy is an example of measurement accuracy information of the present invention.

[0047] (How to correct feature position) Next, a method for correcting the position of a feature will be described. In EKF (Extended Karman Filter) position estimation using road signs and the like using a lidar, the vehicle position is calculated using the difference between the feature measurement value and the feature prediction value, as shown in the above equation (2). As can be seen from equation (2), as the difference value shown in equation (1) increases, the amount of correction to the predicted vehicle position Xe(k) increases. Here, the difference value increases due to one of the following reasons (A) to (C).

[0048] (A) There is an error in the feature measurement value. This applies, for example, when the lidar's measurement accuracy is low, or when the lidar's measurement accuracy is not low but the measurement accuracy is low because a moving object such as another vehicle is present between the feature being measured and the vehicle at the time of measurement (so-called occlusion). (B) The estimated vehicle position used to calculate the feature prediction value is misaligned. This is the case when the error in the predicted vehicle position Xe(k) or the measured updated vehicle position Xp(k) is large. (C) The location coordinates of features in the map data used to calculate feature prediction values ​​are misaligned. This occurs when, for example, a feature sign becomes tilted due to a vehicle collision or a feature structure is removed or moved after the map data is created, changing its position, resulting in the feature's position coordinates in the map data no longer matching the actual feature's position coordinates. In other words, a discrepancy occurs between the feature's position on the map data and its position in the real environment. Therefore, in this case, it is necessary to correct the feature's position coordinates in the map data so that they match the actual feature's position coordinates. In this embodiment, the position of such a feature is corrected using the following method.

[0049] FIG. 6 shows the positional relationship between the estimated vehicle position and features. The coordinate system shown in FIG. 6 is a body coordinate system based on the vehicle, with the x-axis indicating the direction of travel of the vehicle and the y-axis indicating the direction perpendicular to that (the left-right direction of the vehicle). In FIG. 6, the estimated vehicle position has an error range indicated by an error ellipse 40. The error ellipse 40 represents the position estimation accuracy σ in the x-direction. P (x) and y-direction position estimation accuracy σ P (y), where the position estimation accuracy σ P is obtained by transforming the covariance matrix P(k) into the body coordinate system using the Jacobian matrix H(i) according to the following equation (3).

number

[0050] 6, feature predicted values ​​41 are obtained by modeling the measurement process by the LIDAR 2 using the predicted vehicle position Xe(k) and the position vector of the feature registered in the map DB 13. Also, feature measurement values ​​42 are measurement values ​​of the feature by the LIDAR 2 when the position vector of the feature registered in the map DB 13 can be associated with the scan data of the LIDAR 2. The feature measurement values ​​42 have an error range indicated by an error ellipse 43. Generally, the measurement error by the LIDAR increases in proportion to the square of the distance between the LIDAR and the feature to be measured, so by calculating the error ellipse 43 based on the distance from the estimated vehicle position to the feature measurement value, the LIDAR measurement accuracy σ L (σ L (x),σ L The difference value between the feature predicted value 41 and the feature measured value 42 is dx in the x direction and dy in the y direction, as shown in equation (1).

[0051] 7 is a flowchart of the difference value evaluation process performed by the on-board device 10 of the vehicle 1. This process is actually performed by the control unit 15 of the on-board device 10 executing a program prepared in advance. First, the on-board device 10 obtains the difference value (dy, dy) between the feature measurement value 42 and the feature prediction value 41 using the above-mentioned equation (1) (step S11). Next, the on-board device 10 calculates the position estimation accuracy σ from the covariance matrix P(k) using equation (3). P (Step S12). Furthermore, the vehicle-mounted device 10 calculates the lidar measurement accuracy σ based on the distance to the feature. L is calculated (step S13).

[0052] Thus, the difference value (dx, dy) and the position estimation accuracy σ P (σ P (x),σ P (y)), lidar measurement accuracy σ L (σ L (x),σ L Once the estimated vehicle position (y) is obtained, the vehicle-mounted device 10 evaluates the difference value using the position estimation accuracy and the LIDAR measurement accuracy. That is, the vehicle-mounted device 10 determines the validity of the difference between the feature measurement value 42 and the feature prediction value 41 by looking at the ratio of the difference value to the error range 40 of the estimated vehicle position and the LIDAR error range 43. Specifically, the vehicle-mounted device 10 calculates the evaluation values ​​Ex and Ey using the following evaluation formula (4) (step S14).

number

[0053] Next, the vehicle-mounted device 10 compares the evaluation values ​​Ex and Ey with predetermined values ​​(step S15). If at least one of the evaluation values ​​Ex and Ey is greater than the predetermined value (step S15: Yes), the vehicle-mounted device 10 determines that the increase in the difference value is likely due to the above-mentioned (A) or (C), i.e., that (C) is a possibility. Therefore, the vehicle-mounted device 10 obtains, as error information, the date and time information at that time, the estimated vehicle position at that time, the feature ID of the target feature, the difference value (dx, dy), and the position estimation accuracy σ P (σ P (x),σ P(y)), lidar measurement accuracy σ L (σ L (x),σ L (y)) is transmitted to the server device 20 (step S16). Then, the process ends. The reason why the date and time information is included in the error information transmitted to the server device 20 is that it is important that when the actual position coordinates of the feature change due to an environmental change of the feature, etc., the date and time information can be used to know when the change occurred.

[0054] On the other hand, if both the evaluation values ​​Ex and Ey are smaller than the predetermined value (step S15: No), the vehicle-mounted device 10 ends the process. In the case of (B) above, since the denominator of the evaluation formula (4) is large, the evaluation values ​​Ex and Ey do not become large even if the numerators |dx| and |dy| are large. In this way, the difference value evaluation process in the vehicle-mounted device 10 detects that the cause of the large difference value may be (C) above, and error information at that time is transmitted to the server device 20.

[0055] Next, the map correction process performed by the server device 20 will be described. Fig. 8 is a flowchart of the map correction process. This process is actually performed by the control unit 25 of the server device 20 executing a program prepared in advance. First, the server device 20 acquires error information from a plurality of vehicles (step S21). Next, the server device 20 determines whether or not a predetermined number or more pieces of error information have been acquired for the same feature (step S22).

[0056] If the server device 20 has not acquired a predetermined number of pieces of error information for the same feature (step S22: No), the server device 20 determines that the reason why the difference value became large on the vehicle side is (A), that is, the difference value became large by chance due to occlusion or the like, and terminates the processing.

[0057] On the other hand, if more than a predetermined number of pieces of error information have been acquired for the same feature (step S22: Yes), the server device 20 determines that the reason for the large difference value on the vehicle side is (C), that is, the location coordinates of the feature in the map data do not match the actual location coordinates of the feature. Since the difference value is the amount of correction in the vehicle's traveling direction and left / right direction in the body coordinate system shown in FIG. 6, the server device 20 converts the difference values ​​(dx, dy) included in the error information received from the multiple vehicle-mounted devices 10 into deviation amounts in the map coordinate system. Then, the server device 20 determines, by weighted statistical processing, correction difference values ​​that indicate the amount of correction for the feature position in the map data (step S23). Specifically, the server device 20 calculates the position estimation accuracy σ of each vehicle included in the error information received from the vehicle-mounted devices 10. P and lidar measurement accuracy σ L For example, the position estimation accuracy σ P and lidar measurement accuracy σ L The error information with a smaller sum is determined to be highly accurate, and is weighted accordingly. In this way, the server device 20 determines the correction difference value.

[0058] Note that, if the server device 20 has not acquired a predetermined number of pieces of error information for the same feature (step S22: No), it determines that the cause of the large difference value on the vehicle side is (A), and if the server device 20 has acquired a predetermined number of pieces of error information for the same feature (step S22: Yes), it determines that the cause of the large difference value on the vehicle side is (C). However, it is not necessary to determine whether the cause of the large difference value is (A) or (C). In other words, if the server device 20 has not acquired a predetermined number of pieces of error information for the same feature (step S22: No), it may end the processing, and if the server device 20 has acquired a predetermined number of pieces of error information for the same feature (step S22: Yes), it may determine a correction difference value indicating the amount of correction to the feature position in the map data by weighted statistical processing (step S23).

[0059] Next, the server device 20 determines whether the variation in each difference value during the calculation of the correction difference value is equal to or less than a predetermined value (step S24). If the variation in the correction difference value is equal to or less than the predetermined value (step S24: Yes), the server device 20 uses the correction difference value to correct the position coordinates of the feature in the map data (step S25). Note that the correction difference value is the amount of correction in the vehicle's traveling direction and left / right direction in the body coordinate system shown in FIG. 6, so the server device 20 converts these into deviation amounts in the map coordinate system before correcting the position coordinates of the feature in the map data. In this way, the position coordinates of the feature in the map data are corrected based on error information obtained from multiple vehicles. The correction difference value used here is calculated by weighting the difference values ​​included in the error information obtained from multiple on-board devices 10 with the position estimation accuracy and the LIDAR measurement accuracy, so the position coordinates in the map data can be corrected without being affected by the position estimation accuracy and the LIDAR measurement accuracy of each on-board device 10.

[0060] On the other hand, if the variation in the correction difference value is not equal to or less than the predetermined value (step S24: No), it is difficult to accurately determine the correction amount for the position coordinates of the feature in the map data. Therefore, the server device 20 reduces the position accuracy information of the feature in the map data (step S26). That is, the server device 20 adds information indicating that the position accuracy of the feature is low to the feature in the map data. Then, the processing ends. As a result, the vehicle-mounted device 10 knows that the position accuracy of the feature in the map data is low when using the information of the feature, and can take measures such as reducing the weight in the vehicle position estimation process or not using the feature at all. In practice, such features are often measured again using a dedicated measurement vehicle, etc., but by adding information indicating that the position accuracy is low as a temporary measure, it is possible to prevent the vehicle-mounted device 10 in the vehicle 1 from performing driving assistance, autonomous driving, etc. using information with low position accuracy.

[0061] [Second Example] (Vehicle position estimation process) Next, a second embodiment of the process of estimating the vehicle position by the vehicle position estimation unit 17 will be described. The vehicle position estimation unit 17 performs vehicle position estimation by sequentially repeating a prediction step and a measurement update step. As the state estimation filter used in these steps, various filters developed to perform Bayesian estimation can be used, such as an extended Kalman filter, an unscented Kalman filter, and a particle filter. As such, various methods have been proposed for position estimation based on Bayesian estimation. Below, a brief description will be given of vehicle position estimation using an extended Kalman filter as an example.

[0062] FIG. 9 is a diagram showing the state variable vector X in three-dimensional Cartesian coordinates. In the second embodiment, the z-coordinate is taken into consideration. As shown in FIG. 9, the vehicle position on a plane defined on the three-dimensional Cartesian coordinates of xyz is represented by the coordinates "(x, y, z)" and the vehicle's orientation "Ψ." Here, the orientation Ψ is defined as the angle between the vehicle's traveling direction and the x-axis. The coordinates (x, y, z) indicate an absolute position equivalent to, for example, a combination of latitude, longitude, and altitude. Furthermore, because roads have gradients in the front, rear, left, and right directions, the vehicle's orientation also includes a roll angle φ and a pitch angle θ. Therefore, the vehicle's position and orientation can be defined by a total of six variables.

[0063] FIG. 10 is a diagram showing the general relationship between the prediction step and the measurement update step. As shown in FIG. 10, the prediction step and the measurement update step are repeated to sequentially calculate and update the estimated value of the state variable vector X. Here, the state variable vector at the reference time (i.e., the current time) "k" to be calculated is expressed as "Xe(k)" or "Xp(k)". The provisional predicted value predicted in the prediction step is given the subscript "e", and the more accurate estimated value updated in the measurement update step is given the subscript "p". ("State variable vector Xe(k)=(x e (k), y e (k), z e (k), φ e (k), θ e (k), Ψ e (k)) T", "State variable vector Xp(k)=(x p (k), y p (k), z p (k), φ p (k), θ p (k), Ψ p (k)) T " is written as ".)

[0064] In the prediction step, the vehicle position estimation unit 17 calculates the vehicle's moving speed "v" and roll angular velocity "ω" for the state variable vector Xp(k-1) at time k-1 calculated in the immediately preceding measurement update step. x ” and pitch angular velocity “ω y ” and yaw angular velocity “ω z (These will be collectively referred to as "control value u(k)") to calculate a predicted value of the vehicle position at time k (also referred to as "predicted vehicle position") Xe(k). At the same time, the vehicle position estimation unit 17 calculates a covariance matrix Pe(k) corresponding to the error distribution of the predicted vehicle position Xe(k) from the covariance matrix Pp(k-1) at time k-1 calculated in the immediately preceding measurement update step.

[0065] In the measurement update step, the vehicle position estimation unit 17 associates the position vectors of features registered in the map DB 13 with the scan data of the LIDAR 2. When the vehicle position estimation unit 17 has established this association, it acquires the measurement value (referred to as the "feature measurement value") "Z(i)" of the associated feature by the LIDAR 2 and the predicted value (referred to as the "feature predicted value") "Ze(i)" of the feature obtained by modeling the measurement process by the LIDAR 2 using the predicted vehicle position Xe(k) and the position vector of the feature registered in the map DB 13. The feature measurement value Z(i) is a three-dimensional vector obtained by converting the distance and scan angle of the feature measured by the LIDAR 2 at time i into components with axes in the vehicle's traveling direction, lateral direction, and vertical direction. The vehicle position estimation unit 17 then calculates the difference between the feature measurement value Z(i) and the feature predicted value Ze(i) as shown in the following equation (5).

number

[0066] Furthermore, the vehicle position estimation unit 17 multiplies the difference between the feature measurement value Z(i) and the feature prediction value Ze(i) by the Kalman gain "K(k)" and adds this to the predicted vehicle position Xe(k), as shown in the following equation (6), to calculate an updated state variable vector (also called "measurement updated vehicle position") Xp(k).

number

[0067] In the measurement update step, the vehicle position estimation unit 17 calculates a covariance matrix Pp(k) corresponding to the error distribution of the measurement update vehicle position Xp(k) from the prior covariance matrix Pe(k), similar to the prediction step. Parameters such as the Kalman gain K(k) can be calculated in the same manner as in a known vehicle positioning technique using an extended Kalman filter, for example.

[0068] When the vehicle position estimation unit 17 is able to associate the position vectors of multiple features registered in the map DB 13 with the scan data of the LIDAR 2, the vehicle position estimation unit 17 may perform the measurement update step based on any one selected feature measurement value, or may perform the measurement update step multiple times based on all of the associated feature measurement values. When using multiple feature measurement values, the vehicle position estimation unit 17 takes into account that the LIDAR measurement accuracy deteriorates the further a feature is from the LIDAR 2, and decreases the weighting for that feature the longer the distance between the LIDAR 2 and the feature. Generally, the major axis, minor axis, and angle of an error ellipse indicating the expected error range (for example, a range based on a 99% confidence interval) can be calculated from the eigenvalues ​​and eigenvectors of a covariance matrix corresponding to the above-mentioned error distribution.

[0069] In this way, the prediction step and the measurement update step are repeatedly performed, and the predicted vehicle position Xe(k) and the measurement update vehicle position Xp(k) are calculated sequentially, thereby calculating the most probable vehicle position. The vehicle position calculated in this way is called the "estimated vehicle position."

[0070] In the above explanation, the feature measurement value is an example of first distance information of the present invention, the feature prediction value is an example of second distance information of the present invention, the position estimation accuracy is an example of self-position accuracy information of the present invention, and the lidar measurement accuracy is an example of measurement accuracy information of the present invention.

[0071] (How to correct feature position) Next, a method for correcting the position of a feature will be described. In EKF (Extended Karman Filter) position estimation using road signs and the like using a lidar, the vehicle position is calculated using the difference between the feature measurement value and the feature prediction value, as shown in the above equation (6). As can be seen from equation (6), as the difference value shown in equation (5) increases, the amount of correction to the predicted vehicle position Xe(k) increases. Here, the difference value increases due to one of the following reasons (A) to (C).

[0072] (A) There is an error in the feature measurement value. This applies, for example, when the lidar's measurement accuracy is low, or when the lidar's measurement accuracy is not low but the measurement accuracy is low because a moving object such as another vehicle is present between the feature being measured and the vehicle at the time of measurement (so-called occlusion). (B) The estimated vehicle position used to calculate the feature prediction value is misaligned. This is the case when the error in the predicted vehicle position Xe(k) or the measured updated vehicle position Xp(k) is large. (C) The location coordinates of features in the map data used to calculate feature prediction values ​​are misaligned. This occurs when, for example, a feature sign becomes tilted due to a vehicle collision or a feature structure is removed or moved after the map data is created, changing its position, resulting in the feature's position coordinates in the map data no longer matching the actual feature's position coordinates. In other words, a discrepancy occurs between the feature's position on the map data and its position in the real environment. Therefore, in this case, it is necessary to correct the feature's position coordinates in the map data so that they match the actual feature's position coordinates. In this embodiment, the position of such a feature is corrected using the following method.

[0073] FIG. 11 shows the positional relationship between the estimated vehicle position and features when viewed on the xy plane. The coordinate system shown in FIG. 11 is a body coordinate system based on the vehicle, with the x-axis indicating the direction of travel of the vehicle and the y-axis indicating the direction perpendicular to that (the left-right direction of the vehicle). The front side of the paper is the z-axis direction, which indicates the vertical direction of the vehicle. In FIG. 11, the estimated vehicle position has an error range indicated by an error ellipse 40. The error ellipse 40 indicates the position estimation accuracy σ in the x-direction. P (x) and y-direction position estimation accuracy σ P (y) and z-direction position estimation accuracy σ P (z), where the position estimation accuracy σ P is obtained by transforming the covariance matrix P(k) into the body coordinate system using the Jacobian matrix H(i) according to the following equation (7).

number

[0074] 11, feature predicted values ​​41 are obtained by modeling the measurement process by the LIDAR 2 using the predicted vehicle position Xe(k) and the position vector of the feature registered in the map DB 13. Also, feature measurement values ​​42 are measurement values ​​of the feature by the LIDAR 2 when the position vector of the feature registered in the map DB 13 can be associated with the scan data of the LIDAR 2. The feature measurement values ​​42 have an error range indicated by an error ellipse 43. Generally, the measurement error by the LIDAR increases in proportion to the square of the distance between the LIDAR and the feature to be measured, so by calculating the error ellipse 43 based on the distance from the estimated vehicle position to the feature measurement value, the LIDAR measurement accuracy σ L (σ L (x),σ L (y),σ L The difference between the feature predicted value 41 and the feature measured value 42 is dx in the x direction, dy in the y direction, and dz in the z direction, as shown in equation (5).

[0075] 12 is a flowchart of the difference value evaluation process performed by the on-board device 10 of the vehicle 1. This process is actually performed by the control unit 15 of the on-board device 10 executing a program prepared in advance. First, the on-board device 10 acquires the difference values ​​(dy, dy, dz) between the feature measurement values ​​42 and the feature prediction values ​​41 using the above-mentioned equation (5) (step S31). Next, the on-board device 10 calculates the position estimation accuracy σ from the covariance matrix P(k) using equation (7). P (Step S32). Furthermore, the vehicle-mounted device 10 calculates the lidar measurement accuracy σ based on the distance to the feature. L is calculated (step S33).

[0076] Thus, the difference value (dx, dy, dz) and the position estimation accuracy σ P (σ P (x),σ P (y),σ P (z)), lidar measurement accuracy σ L (σ L (x),σ L (y),σ LOnce the estimated vehicle position (z) is obtained, the vehicle-mounted device 10 evaluates the difference value using the position estimation accuracy and the LIDAR measurement accuracy. That is, the vehicle-mounted device 10 determines the validity of the difference between the feature measurement value 42 and the feature prediction value 41 by looking at the ratio of the difference value to the error range 40 of the estimated vehicle position and the LIDAR error range 43. Specifically, the vehicle-mounted device 10 calculates the evaluation values ​​Ex, Ey, and Ez using the following evaluation formula (8) (step S34).

number

[0077] Next, the vehicle-mounted device 10 compares the evaluation values ​​Ex, Ey, and Ez with predetermined values ​​(step S35). If at least one of the evaluation values ​​Ex, Ey, and Ez is greater than the predetermined value (step S35: Yes), the vehicle-mounted device 10 determines that the increase in the difference value is likely due to the above-mentioned (A) or (C), i.e., that (C) is a possibility. Therefore, the vehicle-mounted device 10 obtains, as error information, the date and time information at that time, the estimated vehicle position at that time, the feature ID of the target feature, the difference value (dx, dy, dz), and the position estimation accuracy σ P (σ P (x),σ P (y),σ P (z)), lidar measurement accuracy σ L (σ L (x),σ L (y),σ L (z)) is transmitted to the server device 20 (step S36). Then, the process ends. The reason why the date and time information is included in the error information transmitted to the server device 20 is that it is important that when the actual position coordinates of the feature change due to an environmental change of the feature, etc., the date and time information can be used to know when the change occurred.

[0078] On the other hand, if all of the evaluation values ​​Ex, Ey, and Ez are smaller than the predetermined value (step S35: No), the vehicle-mounted device 10 ends the process. In the case of (B) above, since the denominator of evaluation formula (8) is large, even if the numerators |dx|, |dy|, and |dz| are large, the evaluation values ​​Ex, Ey, and Ez do not become large. In this way, the difference value evaluation process in the vehicle-mounted device 10 detects that the cause of the large difference value may be (C) above, and error information at that time is transmitted to the server device 20.

[0079] Next, the map correction process performed by the server device 20 will be described. Fig. 13 is a flowchart of the map correction process. This process is actually performed by the control unit 25 of the server device 20 executing a program prepared in advance. First, the server device 20 acquires error information from a plurality of vehicles (step S41). Next, the server device 20 determines whether or not a predetermined number or more pieces of error information have been acquired for the same feature (step S42).

[0080] If the server device 20 has not acquired a predetermined number of pieces of error information for the same feature (step S42: No), the server device 20 determines that the reason why the difference value became large on the vehicle side is (A), that is, the difference value became large by chance due to occlusion or the like, and terminates the processing.

[0081] On the other hand, if more than a predetermined number of pieces of error information have been acquired for the same feature (step S42: Yes), the server device 20 determines that the reason for the large difference value on the vehicle side is (C), that is, the location coordinates of the feature in the map data do not match the actual location coordinates of the feature. Since the difference value is the amount of correction in the vehicle's traveling direction, left / right direction, and vertical direction in the body coordinate system shown in FIG. 11, the server device 20 converts the difference values ​​(dx, dy, dz) included in the error information received from the multiple vehicle-mounted devices 10 into deviation amounts in the map coordinate system. Then, the server device 20 determines, by weighted statistical processing, correction difference values ​​that indicate the amount of correction for the feature position in the map data (step S43). Specifically, the server device 20 calculates the position estimation accuracy σ of each vehicle included in the error information received from the vehicle-mounted devices 10.P and lidar measurement accuracy σ L For example, the position estimation accuracy σ P and lidar measurement accuracy σ L The error information with a smaller sum is determined to be highly accurate, and is weighted accordingly. In this way, the server device 20 determines the correction difference value.

[0082] Note that, if the server device 20 has not acquired a predetermined number of pieces of error information for the same feature (step S42: No), it determines that the cause of the large difference value on the vehicle side is (A), and if the server device 20 has acquired a predetermined number of pieces of error information for the same feature (step S42: Yes), it determines that the cause of the large difference value on the vehicle side is (C). However, it is not necessary to determine whether the cause of the large difference value is (A) or (C). That is, if the server device 20 has not acquired a predetermined number of pieces of error information for the same feature (step S42: No), it may end the processing, but if the server device 20 has acquired a predetermined number of pieces of error information for the same feature (step S42: Yes), it may convert the difference values ​​(dx, dy, dz) included in the error information received from the multiple vehicle-mounted units 10 into deviation amounts in the map coordinate system and determine a correction difference value indicating the amount of correction to the feature position in the map data by weighted statistical processing (step S43).

[0083] Next, the server device 20 determines whether the variation in each difference value during the calculation of the correction difference value is equal to or less than a predetermined value (step S44). If the variation in the correction difference value is equal to or less than the predetermined value (step S44: Yes), the server device 20 uses the correction difference value to correct the position coordinates of the feature in the map data (step S45). Note that the correction difference value is the amount of correction in the vehicle's traveling direction, left / right direction, and vertical direction in the body coordinate system shown in FIG. 11 , so the server device 20 converts these into deviation amounts in the map coordinate system before correcting the position coordinates of the feature in the map data. In this way, the position coordinates of the feature in the map data are corrected based on error information obtained from multiple vehicles. The correction difference value used here is calculated by weighting the difference values ​​included in the error information obtained from multiple on-board devices 10 with the position estimation accuracy and the LIDAR measurement accuracy, and therefore the position coordinates in the map data can be corrected without being affected by the position estimation accuracy and the LIDAR measurement accuracy of each on-board device 10.

[0084] On the other hand, if the variation in the correction difference value is not equal to or less than the predetermined value (step S44: No), it is difficult to accurately determine the correction amount for the position coordinates of the feature in the map data. Therefore, the server device 20 reduces the position accuracy information of the feature in the map data (step S46). That is, the server device 20 adds information indicating that the position accuracy of the feature is low to the map data. Then, the processing ends. As a result, the vehicle-mounted device 10 knows that the position accuracy of the feature in the map data is low when using the information of the feature, and can take measures such as reducing the weight in the vehicle position estimation process or not using the feature at all. In practice, such features are often measured again using a dedicated measurement vehicle, etc., but by adding information indicating that the position accuracy is low as a temporary measure, it is possible to prevent the vehicle-mounted device 10 in the vehicle 1 from performing driving assistance, autonomous driving, etc. using information with low position accuracy.

[0085] [Specific example] Next, the effect of the vehicle position estimation process based on the above embodiment will be described.

[0086] Fig. 14(A) is a graph showing the transition of the position in the x direction and the y direction when a vehicle is driven on a driving test course while executing the vehicle position estimation process according to the embodiment. Fig. 14(B) is an enlarged view of the graph for the driving section within frame 91 in Fig. 14(A). In Fig. 14(B), the positions of the features that were the measurement targets are plotted with circles. Fig. 14(C) is a graph showing the transition of the difference value dx in the driving section within frame 91. The difference value dx is close to zero ("0").

[0087] Here, as a simulation, the position of the feature indicated by reference numeral 92 in FIG. 14(B) was intentionally shifted 80 cm forward and the vehicle position estimation process was performed. As a result, the difference value dx was calculated to be 0.749 [m]. FIG. 14(D) shows the transition of the difference value dx obtained in this case in the travel section of frame 91. As indicated by circle 93, at the time when the intentionally shifted feature indicated by reference numeral 92 was detected, the difference value obtained was close to the position shift of that feature.

[0088] The numerical value at the time when the object 92 was detected after shifting its position is the difference value dx=0.749 [m] in the x direction, and the vehicle position estimation accuracy σ p (x)=0.048[m], lidar measurement accuracy σ L (x) = 0.058 [m]. Using these, the evaluation value Ex is calculated using equation (8):

[0089]

number

[0090] As can be seen from the simulation results, the vehicle position estimation process of the embodiment calculates a value close to the positional deviation of the feature. Therefore, if data from multiple vehicles is collected in the server device 20 as described above, it becomes possible to obtain a value close to the actual amount of deviation of the feature's position.

[0091] [Variations] Modifications suitable for the embodiment will be described below. The following modifications may be applied to the embodiment in combination. (Variation 1) In the above embodiment, if the difference between the feature measurement value and the feature prediction value is large, it is determined that the location coordinates of the feature in the map data do not match the actual location coordinates. However, even if the location coordinates of the feature in the map data are correct, the difference value may be large due to the type or shape of the feature. For example, when road signs are used as features, if multiple road signs are lined up as shown in Figure 15, the vehicle-mounted device 10 may detect them as the same feature. Even if the feature prediction value calculated based on the map data distinguishes between two signs, if the feature measurement value cannot distinguish between the two signs, the difference value will be large.

[0092] In such a case, the vehicle 1 equipped with a camera transmits the error information including the image captured by the camera to the server device 20. In this case, the position coordinates of the feature in the map data are correct, so there is no need to correct the position coordinates. However, this feature can be said to be a feature that is prone to errors when used for vehicle position estimation. Therefore, the server device 20 analyzes the captured image using, for example, image recognition by artificial intelligence (AI), and if it determines that the feature is prone to errors, it adds information (for example, a "non-recommended flag") to the feature indicating that it is not recommended to use the feature in position estimation. This allows the on-board device 10 of the vehicle 1 to take measures such as not actively using features with a non-recommended flag added in position estimation processing.

[0093] The same thing can happen when road paint such as white lines is used as a feature. Even beautifully drawn road paint can easily lead to matching errors when compared with lidar detection results. For example, complex white lines such as those shown in FIGS. 16(A) to 16(C) and white lines with road studs 50 such as those shown in FIG. 16(D) require processing to determine the center of the white line from the lidar measurement data. If the lidar detection data is simply identified by reflectance, the white line will be thickened or distorted, making accurate matching impossible.

[0094] Therefore, when the on-board device 10 of the vehicle 1 detects a location where there is a large difference when matching the lidar detection data with the map data, the on-board device 10 includes information indicating this in the error information and transmits this information to the server device 20. The server device 20 can collect error information from multiple vehicles 1 and add information indicating whether the feature (road surface paint) is easy or difficult to match. Furthermore, the on-board device 10 of the vehicle 1 can refer to this information and take measures such as lowering the weight of the feature in position estimation or not using it.

[0095] (Variation 2) In the above embodiment, the on-board device 10 of the vehicle 1 performs the differential value evaluation process for evaluating the differential value, and the server device 20 performs the map correction process. Alternatively, the on-board device 10 may transmit data on all features to the server device 20, and the server device 20 may perform the differential value evaluation process and the map correction process. Specifically, the on-board device 10 transmits error information on all features (including date and time information, estimated vehicle position, feature ID, differential value, position estimation accuracy, and LIDAR measurement accuracy) to the server device 20 without performing evaluation using the evaluation values ​​Ex, Ey, and Ez. The server device 20 may first calculate the evaluation values ​​Ex, Ey, and Ez using the received error information, and then perform the map correction process shown in FIG. 8 or 13 for error information in which these values ​​exceed a predetermined value.

[0096] (Variation 3) Conversely, the difference value evaluation process and the map correction process may be performed by the onboard device 10 of the vehicle 1. In this case, the onboard device 10 first executes the difference value evaluation process shown in FIG. 7 or FIG. 12 to generate error information and store it in the storage unit 12. When the vehicle 1 travels through the same location multiple times, multiple pieces of error information for the same feature are obtained. Therefore, the onboard device 10 performs the map correction process shown in FIG. 8 or FIG. 13 using multiple pieces of error information for the same feature stored in the storage unit 12. That is, instead of using error information acquired from multiple vehicles, multiple pieces of error information acquired by multiple travels of the vehicle are used. In this way, the onboard device 10 corrects the feature stored in the map DB 13 using the correction difference value or performs a process to reduce the position accuracy information for the feature. Furthermore, if necessary, the map data after such correction may be transmitted to the server device 20.

[0097] (Variation 4) Regarding the evaluation formula, the following formula (10) may be used instead of the above formula (8).

number

[0098] (Variation 5) The position of the object may be calculated based on the measurement by adding the measured value of the distance to the object to the estimated vehicle position. The calculated position of the object based on the measurement may then be compared with the position of the object in the map information to determine whether there is a deviation.

[0099] (Variation 6) In the above embodiment, the server device 20 collects error information from the vehicle-mounted device 10. However, if a vehicle measures a specific feature multiple times, the vehicle-mounted device 10 may calculate a weighted average and variance for the difference values ​​(dy, dy) for the multiple measurements, and transmit the error information to the server device 20 if the variance value is small.

[0100] Specifically, if the evaluation values ​​Ex and Ey are greater than the predetermined values ​​and the reason for the large difference value does not fall under "(B) The estimated vehicle position used to calculate the feature predicted value is shifted," the vehicle-mounted device 10 calculates the difference value (dx, dy) calculated N times as the position estimation accuracy σ P (σ P (x),σ P (y)) and lidar measurement accuracy σ L (σ L (x),σ L The weighted mean and variance are calculated using (dy, dy). That is, the weighting of (dy, dy) is increased when the accuracy is high, and decreased when the accuracy is low.

[0101] The weighted average of the difference values ​​dx is calculated by the following formula (11): The weighted average of the difference values ​​dy is calculated by a formula for the difference values ​​dy similar to formula (11), that is, a formula in which "x" in formula (11) is replaced with "y".

number

[0102] The variance of the difference value dx is calculated by the following equation (12): The variance of the difference value dy is calculated by an equation for the difference value dy similar to equation (12), that is, an equation in which "x" in equation (12) is replaced with "y".

number

[0103] If the variance in this calculation result is large, it is clear that there is a lot of variation in the difference values, so it is determined that the cause of the large difference values ​​is likely to be "(A) there is an error in the feature measurement values," and the vehicle-mounted device 10 terminates the process without transmitting any error information to the server device 20. On the other hand, if the variance is small, it is clear that there is little variation in the difference values, so it is determined that the cause of the large difference values ​​is likely to be "(C) the position coordinates of the feature in the map data used to calculate the feature predicted value are misaligned," and the vehicle-mounted device 10 transmits the weighted average value of the difference values ​​(dy, dy) and the accuracy of the weighted average value to the server device 20.

[0104] Here, the precision of the weighted average value of the difference value dx is given by the following formula (13): Furthermore, the precision of the weighted average value of the difference value dy is calculated by a formula for the difference value dy similar to formula (13), i.e., a formula in which "x" in formula (13) is replaced with "y".

number

[0105] The vehicle-mounted device 10 calculates the difference values ​​(dy, dy) for N times, the position estimation accuracy σ P (σ P (x),σ P (y)) and lidar measurement accuracy σ L (σ L (x),σ L (y)) may be transmitted to the server device 20, and the server device may use the above formula to calculate the weighted mean and variance, determine (A) and (C) based on the variance, and calculate the accuracy of the weighted mean value.

[0106] By performing such processing in the vehicle-mounted device 10, only data that is likely to be caused by "(C) the location coordinates of features in the map data used to calculate the feature prediction value are shifted" is transmitted to the server device 20, thereby reducing the frequency of data transmission from the vehicle-mounted device 10 to the server device 20 and reducing the processing on the server device 20 side.

[0107] Furthermore, the server device 20 may decide whether or not to perform the above-described processing of the on-vehicle device 10, and if the processing is to be performed, the server device 20 may transmit request information to the on-vehicle device 10 to perform the above-described processing. Specifically, the server device 20 determines whether or not it is necessary to reconcile the positions of the features stored in the map DB 13 based on the last update date, the accuracy of the map information, etc. If it is necessary to reconcile the positions of the features, the server device 20 does not transmit request information to perform the above-described processing, and instead sets the on-vehicle device 10 to transmit error information to the server device 20. On the other hand, if it is not necessary to reconcile the positions of the features, the server device 20 transmits request information to perform the above-described processing to the on-vehicle device 10. This makes it possible to perform processing according to the need for reconciliation for each feature. [Explanation of symbols]

[0108] 1 Onboard device 2 Rider 3 Gyro sensor 4 Vehicle speed sensor 5 GPS receiver 10 Onboard equipment 13 Map DB 15, 25 Control section 20 Server device 23 Map DB

Claims

1. a first acquisition unit that acquires first distance information that indicates a distance from a moving body to an object, the first distance information being measured by a measurement unit, and second distance information that indicates a distance from a position of the moving body to a position of the object in map information, the second distance information being estimated by an estimation unit; a second acquisition unit that acquires measurement accuracy information indicating the measurement accuracy of the position of the object by the measurement unit and estimation accuracy information indicating the estimation accuracy of the position of the moving body by the estimation unit; a determination unit that determines whether or not there is a deviation of a predetermined value or more between the position of the object in the map information and the position of the object in the real environment, based on a difference value between the distances indicated by the first distance information and the second distance information, the measurement accuracy information, and the estimated accuracy information; and A determination device comprising:

2. a third acquisition unit that acquires, from the determination unit, error information including the difference value, the estimated accuracy information, and the measurement accuracy information for the object determined to have a deviation equal to or greater than the predetermined value; a correction unit that, when the third acquisition unit acquires a predetermined number of pieces of error information for the same object, determines a correction difference value indicating a correction amount for the position of the object in the map information by statistical processing of difference values ​​included in the acquired error information, and corrects the position of the object in the map information using the correction difference value; The determination device according to claim 1 , further comprising:

3. the first distance information and the second distance information are vectors in a moving body coordinate system based on the moving body, and the difference value is a difference value vector in the moving body coordinate system; 3. The determination device according to claim 2, wherein the correction unit performs the statistical processing by converting the difference value vector from a moving body coordinate system to a map coordinate system.

4. The determination device described in claim 2, characterized in that when the third acquisition unit acquires error information from a predetermined number or more vehicles for the same object, and the variation in the difference values ​​included in the acquired error information is a predetermined value or more, the correction unit lowers the level of the position accuracy information assigned to the object in the map information.

5. The determination device is mounted on a vehicle, The determination device according to claim 1, further comprising a transmitting unit that transmits error information including the identification information of the object, the difference value, the estimated accuracy information, and the measurement accuracy information to an external server when the determination unit determines that there is a deviation greater than the predetermined value.

6. The determination device described in claim 5, characterized in that when multiple pieces of error information are obtained for the same object, if a variance value based on the multiple pieces of error information is less than a predetermined value, the transmission unit transmits the multiple pieces of error information or information obtained using the multiple pieces of error information to the external server.

7. 6. The determination device according to claim 2, wherein the error information includes date and time information.

8. the measurement unit and the estimation unit are mounted on a vehicle, The determination device is installed in a server, The determination device according to claim 1 , wherein the first acquisition unit acquires the first distance information and the second distance information from the vehicle.

9. A determination method executed by a determination device, a first acquisition step of acquiring first distance information indicating a distance from a moving body to an object, the first distance information being measured by a measurement unit, and second distance information indicating a distance from a position of the moving body to a position of the object in map information, the second distance information being estimated by an estimation unit; a second acquisition step of acquiring measurement accuracy information indicating the measurement accuracy of the position of the object by the measurement unit and estimation accuracy information indicating the estimation accuracy of the position of the moving body by the estimation unit; a determining step of determining whether or not there is a deviation of a predetermined value or more between the position of the object in the map information and the position of the object in the real environment, based on a difference value between the distances indicated by the first distance information and the second distance information, the measurement accuracy information, and the estimated accuracy information; A determination method comprising:

10. A program executed by a determination device including a computer, a first acquisition unit that acquires first distance information that indicates a distance from a moving body to an object, the first distance information being measured by a measurement unit, and second distance information that indicates a distance from a position of the moving body to a position of the object in map information, the second distance information being estimated by an estimation unit; a second acquisition unit that acquires measurement accuracy information indicating the measurement accuracy of the position of the object by the measurement unit and estimation accuracy information indicating the estimation accuracy of the position of the moving body by the estimation unit; a determination unit that determines whether or not there is a deviation of a predetermined value or more between the position of the object in the map information and the position of the object in the real environment, based on a difference value between the distances indicated by the first distance information and the second distance information, the measurement accuracy information, and the estimated accuracy information; A program causing the computer to function as a

11. A storage medium storing the program according to claim 10.

12. a first acquisition unit that acquires first distance information that indicates a distance from a moving body to an object, the first distance information being measured by a measurement unit, and second distance information that indicates a distance from a position of the moving body to a position of the object in map information, the second distance information being estimated by an estimation unit; a second acquisition unit that acquires measurement accuracy information indicating the measurement accuracy of the position of the object by the measurement unit and estimation accuracy information indicating the estimation accuracy of the position of the moving body by the estimation unit; a determination unit that determines whether or not there is a deviation of a predetermined value or more between the position of the object in the map information and the position of the object based on the first distance, based on a difference value between the distances indicated by the first distance information and the second distance information, the measurement accuracy information, and the estimated accuracy information; A determination device comprising:

Images