JP2018116014A - Own vehicle position estimation device, control method, program and storage medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an own vehicle position estimation device capable of estimating a highly accurate own vehicle position by utilizing map data.SOLUTION: A vehicle-mounted device detects point group data representing a shape of a feature existing around a vehicle based on output of a rider. Moreover, the vehicle-mounted device refers to a feature DB21 and generates polygon data indicating the shape of the feature recognized from the outline own vehicle position based on output of a GPS receiver or the like. Then, the vehicle-mounted device estimates a highly accurate own vehicle position based on the point group data and the polygon data.SELECTED DRAWING: Figure 4

Description

  The present invention relates to a position estimation technique using three-dimensional data of a feature.

  Conventionally, a technique for generating three-dimensional data of a feature such as a building is known. For example, Patent Document 1 discloses a technique for analyzing raster format image data such as satellite photographs and aerial photographs to automatically recognize sections such as roads, rivers, and farm fields, and outputting section information as vector format image data. Is disclosed.

JP 2014-194722 A

  In general, the vehicle position based on the output of a GPS receiver or a self-supporting positioning sensor may deviate from the actual vehicle position, and such a displacement is an obstacle to the execution of accurate route guidance and automatic driving control. Become. On the other hand, in Patent Document 1, since three-dimensional data of a feature is automatically generated, map data including the three-dimensional data of the feature can be generated by applying such a technique to map maintenance. Then, this invention makes it the main objective to provide the own vehicle position estimation apparatus which can estimate the own vehicle position with high precision using map data.

  The invention according to claim 1 is an own vehicle position estimating device, wherein a detecting means for detecting a first shape which is a shape of a feature existing around the vehicle, and an approximate own vehicle position of the vehicle are acquired. An acquisition unit; a generation unit that refers to a feature database storing the position and shape of the feature, and generates a second shape that is a shape of the feature recognized from the approximate vehicle position; and the first shape and Estimation means for estimating a high-accuracy vehicle position of the vehicle based on the second shape.

  The invention according to claim 9 is a control method executed by the own vehicle position estimating device, wherein a detection step of detecting a first shape which is a shape of a feature existing around the vehicle, and a rough self of the vehicle. With reference to the acquisition step of acquiring the vehicle position and the feature database storing the position and shape of the feature, the features recognized from the approximate own vehicle position or a candidate position that is a peripheral position of the approximate own vehicle position. A generating step of generating a second shape that is a shape, and an estimating step of estimating a high-accuracy vehicle position of the vehicle based on the first shape and the second shape.

  The invention according to claim 10 is a program executed by a computer, and detects a first shape which is a shape of a feature existing around the vehicle, and acquires a rough own vehicle position of the vehicle. A second feature shape that is recognized from an acquisition means and a feature database that stores the position and shape of the feature, and is recognized from the approximate vehicle position or a candidate position that is a peripheral position of the approximate vehicle position; The computer is caused to function as a generation unit that generates a shape, and an estimation unit that estimates the vehicle position of the vehicle with high accuracy based on the first shape and the second shape.

It is a schematic structure of the own vehicle position estimation system. The block configuration of the in-vehicle device is shown. An example of the data structure of feature DB is shown. It is a flowchart which shows the outline | summary of the own vehicle position estimation process. It is a flowchart which shows a highly accurate own vehicle position determination process. It is a bird's-eye view showing the periphery of the own vehicle position. It is a bird's-eye view showing the outline of the own vehicle position. The relative positional relationship of the irradiated site when the point cloud data and polygon data are expressed in the same coordinate system by coordinate transformation based on the approximate vehicle position is shown. (A) An example of setting candidate positions is shown. (B) The relative positional relationship of the irradiated parts when the point cloud data and the polygon data are represented in the same coordinate system by coordinate conversion with the candidate position P5 as a reference is shown. An example of the data structure of feature DB referred in a modification is shown.

  According to a preferred embodiment of the present invention, there is provided an own vehicle position estimating device for detecting a first shape which is a shape of a feature existing around a vehicle, and an approximate own vehicle position of the vehicle. An acquisition means for acquiring, a generation means for generating a second shape which is a shape of the feature recognized from the approximate vehicle position, with reference to a feature database storing the position and shape of the feature, and the first Estimation means for estimating a high-accuracy host vehicle position of the vehicle based on the shape and the second shape.

  The host vehicle position estimation apparatus includes a detection unit, an acquisition unit, a generation unit, and an estimation unit. The detecting means detects a first shape that is a shape of a feature existing around the vehicle. The acquisition means acquires a rough vehicle position of the vehicle. Here, the “approximate own vehicle position” refers to a provisional and approximate own vehicle position until a highly accurate own vehicle position to be described later is estimated. The generation means refers to a feature database that stores the position and shape of the feature, and generates a second shape that is the shape of the feature recognized from the approximate vehicle position. The estimating means estimates the vehicle position of the vehicle with high accuracy based on the first shape and the second shape. In this aspect, the host vehicle position estimation device accurately determines the host vehicle position based on the shape of the feature detected by the detection unit and the shape of the feature recognized from the feature database and the approximate host vehicle position. Can be estimated.

  In one aspect of the host vehicle position estimation device, the acquisition unit acquires a direction of the vehicle, and the detection unit detects a shape of a feature existing in the direction from the vehicle as the first shape, The generation unit generates, as the second shape, a shape of a feature existing in the direction from the approximate vehicle position or a candidate position set around the approximate vehicle position. In general, the shape that can be detected from the vehicle is limited to the front side when observed from the vehicle, and the back side cannot be detected from the vehicle. Therefore, in this aspect, the vehicle position estimation device preferably matches the first shape detected by the detecting means by generating the shape of the feature that can be detected from the vehicle as the second shape based on the direction of the vehicle. Can be made.

  In another aspect of the host vehicle position estimation device, the estimation unit estimates the approximate host vehicle position as the high-precision host vehicle position when the first shape and the second shape match, and When the first shape does not match the second shape, the shape of the feature recognized from the candidate positions set around the approximate vehicle position is the second shape, and the second shape is the first shape. Is estimated as a high-accuracy vehicle position. According to this aspect, the host vehicle position estimation apparatus can preferably estimate the high-precision host vehicle position.

  In another aspect of the host vehicle position estimation device, the generation unit generates a second shape for a plurality of candidate positions, and the estimation unit is similar to the first shape among a plurality of second shapes. The second shape having the maximum degree and not less than the predetermined threshold value is determined as the second shape that matches the first shape. According to this aspect, the host vehicle position estimation device can suitably select a candidate position closest to the actual host vehicle position as a high-accuracy host vehicle position from a plurality of set candidate positions.

  In another aspect of the host vehicle position estimation device, the generation unit recognizes a feature recognized from the approximate vehicle position or a candidate position of the high-accuracy host vehicle position set around the vehicle position. Among them, a shape of a feature having a predetermined high priority is generated as the second shape. With this aspect, the host vehicle position estimation device reduces the processing load by generating the second shape only for features suitable for estimating the host vehicle position with high accuracy, for example, and reduces the processing load. It is possible to improve the estimation accuracy. Preferably, the high priority feature includes at least one of a feature having a height equal to or higher than a predetermined value and a public feature.

  In another aspect of the vehicle position estimation device, the detection means generates point cloud data of a feature by an external sensor, and a shape specified based on the point cloud data is the first shape. According to this aspect, the host vehicle position estimation device can preferably detect the shape of the feature existing around the vehicle.

  In another aspect of the vehicle position estimation device, the acquisition unit is acquired by a GPS and / or a self-supporting sensor. According to this aspect, the own vehicle position estimation device can appropriately recognize the approximate own vehicle position that is the own vehicle position necessary to generate the second shape.

  According to another preferred embodiment of the present invention, there is provided a control method executed by the vehicle position estimation device, wherein a detection step of detecting a first shape which is a shape of a feature existing around the vehicle, With reference to the acquisition step of acquiring the approximate vehicle position of the vehicle and the feature database storing the position and shape of the feature, the vehicle is recognized from the approximate vehicle position or a candidate position that is a peripheral position of the approximate vehicle position. A generating step of generating a second shape that is a shape of the target feature, and an estimating step of estimating a high-accuracy vehicle position of the vehicle based on the first shape and the second shape. By executing this control method, the host vehicle position estimation device can detect the host vehicle based on the shape of the feature detected by the detection means and the shape of the feature recognized from the feature database and the approximate host vehicle position. The position can be estimated with high accuracy.

  According to another preferred embodiment of the present invention, a program executed by a computer, detecting means for detecting a first shape, which is a shape of a feature existing around a vehicle, and the vehicle itself The shape of the feature recognized from the approximate own vehicle position or a candidate position that is a peripheral position of the approximate own vehicle position with reference to an acquisition means for acquiring a position and a feature database storing the position and shape of the feature Based on the first shape and the second shape, the computer is caused to function as an estimation means for estimating the vehicle position of the vehicle with high accuracy based on the first shape and the second shape. By executing this program, the computer can accurately determine the vehicle position based on the shape of the feature detected by the detection means and the shape of the feature recognized from the feature database and the vehicle position. Can be estimated. Preferably, the program is stored in a storage medium.

  Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings.

[Outline of the vehicle position estimation system]
FIG. 1 is a schematic configuration of the vehicle position estimation system according to the present embodiment. The own vehicle position estimation system includes a vehicle traveling on a road and an in-vehicle device 1 that moves together with the vehicle.

  The in-vehicle device 1 includes a lidar (Light Detection and Ranging or Laser Illuminated Detection And Ranging) 30 installed in the vehicle. And the vehicle equipment 1 estimates a self-vehicle position with high precision based on the point cloud data regarding the feature around the vehicle which the rider 30 outputs. The in-vehicle device 1 is an example of the “own vehicle position estimation device” in the present invention.

  The lidar 30 emits a pulse laser in a predetermined angular range in the horizontal direction and the vertical direction, thereby discretely measuring the distance to an object existing in the outside world, and a three-dimensional point indicating the position of the object Generate group data. In this case, the lidar 30 includes an emitting unit that emits a pulse laser while changing an irradiation direction, a light receiving unit that receives reflected light (scattered light) of the irradiated pulse laser, and a point group based on a light reception signal output from the light receiving unit. And an output unit for outputting data. The point cloud data is generated based on the irradiation direction corresponding to the pulse laser received by the light receiving unit and the response delay time of the pulse laser specified based on the light reception signal. The lidar 30 is an example of the “external sensor” in the present invention. In addition, instead of performing scanning in both the horizontal direction and the vertical direction, the lidar 30 may tilt the emission direction of the pulse laser from the horizontal direction so that the scanning surface is inclined with respect to the horizontal direction.

  FIG. 2 is a block diagram illustrating a functional configuration of the in-vehicle device 1. The in-vehicle device 1 mainly includes a communication unit 11, a storage unit 12, a sensor unit 13, an input unit 14, a control unit 15, and an output unit 16. Each of these elements is connected to each other via a bus line.

  The communication unit 11 performs data communication with other devices based on the control of the control unit 15. The storage unit 12 stores a program executed by the control unit 15 and information necessary for the control unit 15 to execute a predetermined process.

  In the present embodiment, the storage unit 12 stores map data 20 including the feature DB 21. The feature DB 21 is a database related to features existing around the road. FIG. 3 shows a schematic data structure of the feature DB 21. In the example of FIG. 3, the feature DB 21 includes, for each feature existing around the road, a “feature ID” that is identification information of the feature, a “feature type” that represents the type of the feature, “Position information” indicating the absolute position of the feature is associated with “polygon data”. Here, “polygon data” is data representing the outer shape of a feature by polygonal plane data. The feature DB 21 is an example of the “feature database” in the present invention. In addition to the feature DB 21, the map data 20 stores road data and facility information used for route guidance, automatic driving control, and the like.

  With reference to FIG. 2 again, each element of the vehicle-mounted device 1 will be described. The sensor unit 13 includes an internal sensor for detecting the state of the vehicle and an external sensor for recognizing the surrounding environment of the vehicle. In addition to the lidar 30 described above, a camera 31 for capturing a scene outside the vehicle, and a GPS receiver 32 and a self-supporting positioning sensor 33 such as a gyro sensor or a speed sensor. In the present embodiment, the outputs of the GPS receiver 32 and the self-supporting positioning sensor 33 are used to specify the current position of the vehicle and the traveling direction (that is, the direction of the vehicle). Hereinafter, the position of the vehicle specified based on the output of the GPS receiver 32 and / or the self-supporting positioning sensor 33 is also referred to as “substantially own vehicle position”.

  The input unit 14 is a button for operation by the user, a touch panel, a remote controller, a voice input device, or the like. The output unit 16 is, for example, a display or a speaker that performs output based on the control of the control unit 15.

  The control unit 15 includes a CPU that executes a program, and controls the entire vehicle-mounted device 1. In the present embodiment, the control unit 15 compares the point cloud data output from the rider 30 with the polygon data extracted from the feature DB 21, thereby determining the approximate vehicle position specified from the output of the GPS receiver 32 or the like. Is estimated as necessary with a high-accuracy vehicle position (also referred to as “high-accuracy vehicle position”). The control unit 15 is an example of a “detection unit”, “acquisition unit”, “generation unit”, “estimation unit”, and a computer that executes the program according to the present invention.

[Vehicle position estimation processing]
Next, the own vehicle position estimation process executed by the control unit 15 will be described. Schematically, the control unit 15 compares the point cloud data output from the lidar 30 and the polygon data in the feature DB 21 into the same coordinate system for features existing within the measurement range of the lidar 30 and compares them. Thus, the high-precision vehicle position is estimated.

(1) Process Overview FIG. 4 is a flowchart showing an overview of the vehicle position estimation process. The control unit 15 repeatedly executes the process of the flowchart shown in FIG.

  First, the control unit 15 acquires the point cloud data output by the lidar 30 and detects the approximate vehicle position and the traveling direction of the vehicle based on the output of the GPS receiver 32 and / or the independent positioning sensor 33 ( Step S101).

  Next, the control unit 15 estimates the measurement range of the rider 30 based on the approximate vehicle position and traveling direction of the vehicle detected in step S101 (step S102). For example, the storage unit 12 stores in advance information on the measurement range of the rider 30 relative to the traveling direction of the vehicle and information on the maximum distance measurement distance of the rider 30. Then, the control unit 15 recognizes the measurement range in the absolute coordinates (for example, latitude and longitude) of the rider 30 from the approximate vehicle position and traveling direction of the vehicle based on these pieces of information stored in the storage unit 12.

  Next, the control unit 15 refers to the feature DB 21 and extracts polygon data of features existing within the measurement range of the lidar 30 estimated in step S102 (step S103). In this case, the control unit 15 specifies features existing within the measurement range of the rider 30 with reference to the position information of the feature DB 21 and extracts polygon data of the specified features. In this case, the control unit 15 appropriately converts the coordinate system of the polygon data into an absolute three-dimensional coordinate system (for example, latitude, longitude, altitude) based on the position information described above.

  Next, the control unit 15 is a coordinate system between the polygon data of the feature based on the feature DB 21 and the point cloud data output by the lidar 30 based on the approximate vehicle position detected in step S101 and the traveling direction of the vehicle. Are unified (step S104). In the first example, the control unit 15 converts the polygon data expressed in the absolute coordinate system into the same relative coordinate system as the point cloud data output by the lidar 30. In this case, the control unit 15 converts the polygon data of the feature based on the feature DB 21 into a relative coordinate system based on the position and the traveling direction of the vehicle based on the approximate vehicle position and the traveling direction. In the second example, the control unit 15 converts the point group data in the relative coordinate system output from the lidar 30 into the same absolute coordinate system as the polygon data based on the feature DB 21. In this case, the control unit 15 converts the point cloud data output from the lidar 30 into an absolute coordinate system using absolute coordinates (for example, latitude, longitude, altitude) based on the approximate vehicle position and the traveling direction. .

  Next, the control unit 15 collates (matches) the point cloud data after unifying the coordinate system with polygon data based on the map DB 21 (step S105). At this time, preferably, the control unit 15 extracts only polygon data within a range in which the pulse laser can be irradiated from the vehicle from among the polygon data registered in the feature DB 21, and the extracted polygon data and the above point cloud data are extracted. It is good to check with. This extraction process will be described in detail in the section “(3) Specific Example”. Further, the control unit 15 may convert the point cloud data into polygon data, and collate the converted polygon data with polygon data based on the map DB 21. Hereinafter, for convenience of explanation, unless otherwise specified, the polygon data refers to polygon data based on the map DB 21.

  Here, in both cases of the first and second examples described in step S104, coordinate conversion is performed using the approximate vehicle position. Therefore, when the approximate vehicle position is different from the actual vehicle position, the point cloud data and the polygon data do not match. The shape indicated by the point cloud data to be collated in step S105 is an example of the “first shape” in the present invention, and the shape indicated by the polygon data to be collated in step S105 is an example of the “second shape” in the present invention.

  If the control unit 15 collates the point cloud data with the polygon data and determines that they match (step S106; Yes), the approximate vehicle position is directly used as the high-accuracy vehicle position (step S107). . For example, in this case, the control unit 15 calculates the similarity between the point cloud data and the polygon data based on a known three-dimensional pattern matching technique, and when the similarity is equal to or greater than a predetermined threshold, It is determined that the data matches the polygon data. In this case, the control unit 15 regards the approximate vehicle position as the actual vehicle position, and sets the approximate vehicle position as the high-accuracy vehicle position.

  On the other hand, as a result of collating the point cloud data and the polygon data and determining that they do not match (step S106; No), the control unit 15 determines that the approximate vehicle position is deviated from the actual vehicle position. To do. Therefore, in this case, the control unit 15 executes a high-accuracy vehicle position determination process that is a process of setting the high-accuracy vehicle position at a position different from the approximate vehicle position (step S108).

(2) High-Accuracy Own Vehicle Position Determination Process FIG. 5 is a flowchart showing an example of the high-accuracy own vehicle position determination process executed in step S108.

  First, the control unit 15 sets a position (also referred to as “candidate position”) that is a candidate for a high-accuracy own-vehicle position around the approximate own-vehicle position (step S201). In this case, for example, the control unit 15 extracts a predetermined number of positions on the road where the approximate vehicle position exists and the road connected to the road based on a predetermined rule, and sets the extracted positions as candidate positions.

  Next, the control unit 15 unifies the coordinate system of the polygon data extracted in step S103 and the point cloud data output by the lidar 30 with each candidate position as a reference (step S202). In this case, the control unit 15 may execute the process of step S202 according to either the first example or the second example described in step S104.

  Then, the control unit 15 collates the point cloud data and the polygon data after unifying the coordinate system for each candidate position (step S203). Here, the control unit 15 calculates the similarity between the point cloud data and the polygon data for each candidate position based on a known three-dimensional pattern matching method. Then, the control unit 15 sets, as the high-accuracy own vehicle position, a candidate position having a maximum similarity and having a similarity equal to or greater than a predetermined threshold among the candidate positions (step S204). Note that if the maximum similarity among the calculated similarities is less than the predetermined threshold, the control unit 15 may regard any of the candidate positions set in step S201 as a high-accuracy vehicle position. It is determined that it cannot be performed, a candidate position different from the already set candidate positions is newly set, and steps S202 to S204 are executed again.

  Note that the high-accuracy host vehicle position determination process executed in step S108 is not limited to this. For example, the control unit 15 detects the amount of positional deviation between the point cloud data and the polygon data in the collating process in step S105, and corrects the approximate own vehicle position based on the detected amount of deviation, thereby high-accuracy own vehicle. The position may be calculated. In another example, the control unit 15 stores the point cloud data output by the rider 30 and the approximate vehicle position detected at that time in the storage unit 12 for at least a predetermined time and is stored in the storage unit 12. The point cloud data at each position and the polygon data based on the feature DB 21 are collated. And the control part 15 considers the approximate own vehicle position from which the point cloud data which correspond to the polygon data based on feature DB21 were output as a highly accurate own vehicle position.

(3) Specific Example FIG. 6 is an overhead view showing the vicinity of the vehicle position. In the example of FIG. 6, the vehicle exists at the position “P1” on the road 36, and the surrounding features 20 to 24 (20, 21, 24 are buildings, 22 are utility poles, and 23 are post boxes). doing. Here, the arrow 25 indicates the traveling direction of the vehicle, and the alternate long and short dash line “R1” indicates the measurement range of the rider 30.

  Here, the point cloud data output by the lidar 30 is generated for the irradiated object of the pulse laser emitted by the lidar 30 from the own vehicle position P1 within the measurement range R1. In the example of FIG. 6, point cloud data is generated for irradiated portions 40 to 44 (see thick broken lines) that are within the measurement range R1 and are the surfaces of the features 20 to 24 irradiated with the pulse laser. . The irradiated portions 40 to 44 are portions where the pulse laser emitted from the vehicle position P1 arrives without being shielded by other features and the like, for example, by simulation using a known three-dimensional computer graphics technique Identified. And the control part 15 acquires the point group data of the irradiated parts 40-44 from the rider 30 in step S101 of FIG.

  FIG. 7 is a bird's-eye view showing the vicinity of the vehicle position. The position “P2” indicates the approximate vehicle position detected in step S101 in FIG. 4, and the arrow 26 indicates the traveling direction of the vehicle detected in step S101.

  In the example of FIG. 7, the approximate own vehicle position P2 is shifted rearward from the actual own vehicle position P1. In this case, the control unit 15 estimates the measurement range “R2” of the rider 30 based on the approximate vehicle position P2 and the like according to step S102, and according to step S103, the features 20, 22, and 23 existing within the measurement range R2. Polygon data is extracted from the feature DB 21. Further, in the example of FIG. 7, the control unit 15 is within the measurement range R2, and assuming that the pulse laser is emitted from the approximate vehicle position P2, the part of the feature irradiated with the pulse laser (“target” Polygon data corresponding to “part” is also extracted. Specifically, the control unit 15 extracts polygon data corresponding to the target parts 50, 52, and 53 indicated by bold lines from the polygon data of the features 20, 22, and 23.

  FIG. 8 shows the relative relationship between the irradiated portions 40 to 44 and the target portions 50, 52, and 53 when the point cloud data and the polygon data are represented by the same coordinate system by the coordinate transformation in step S 104 based on the approximate vehicle position P 2. Indicates the positional relationship.

  As shown in FIG. 8, the control unit 15 assumes that the approximate vehicle position P2 is equal to the own vehicle position P1, and based on the approximate own vehicle position P2 and the detected traveling direction of the vehicle, the measurement range R1 and the measurement range R2 And match. Here, the rough own vehicle position P2 detected in step S101 of FIG. 4 is shifted rearward from the actual own vehicle position P1, as shown in FIG. 7, and therefore corresponds to the same feature 20. The irradiated site 40 and the target site 50 are not matched. Similarly, the irradiated portions 42 and 52 corresponding to the same feature 22 and the target portions 43 and 53 corresponding to the feature 23 do not match. Therefore, in this case, the control unit 15 determines in step S106 that the point cloud data and the polygon data do not match as a result of the collation in step S105, and performs the high-accuracy vehicle position determination process in step S108.

  FIG. 9A shows a setting example of candidate positions set in step S201 of the high-accuracy own vehicle position determination process shown in FIG. In FIG. 9 (A), the control unit 15 sets the approximate vehicle position P2 to 4 at approximately equal intervals on the road 36 where the approximate vehicle position P2 exists and the road 37 extending in the same direction as the road 36. Two candidate positions “P3” to “P6” are set. The control unit 15 does not set candidate positions for the other roads 38 and 39 connected to the road 36 because the roads cannot travel along the detected traveling direction of the vehicle.

  Then, the control unit 15 unifies the coordinate system of the polygon data and the point cloud data based on each of the set candidate positions P3 to P6 based on step S202, and performs the matching process in step S203.

  FIG. 9B shows the relative positional relationship between the irradiated portions 40 to 44 and the target portions 50 to 54 when the point cloud data and the polygon data are represented by the same coordinate system by the coordinate conversion based on the candidate position P5. Indicates. In this case, the control unit 15 assumes that the candidate position P5 is equal to the host vehicle position P1, and matches the measurement range “R5” of the rider 30 when the candidate position P5 is the host vehicle position with the measurement range R1. Coordinate conversion is performed. In this case, the irradiated portions 40 to 44 and the irradiated portions 50 to 54 corresponding to the same feature 20 to 24 have substantially the same position and the same shape, respectively. Then, as a result of the collation in step S203, the control unit 15 determines that the similarity corresponding to the candidate position P5 is higher than the similarities of the other candidate positions P3, P4, and P6 and is equal to or greater than a predetermined threshold. The candidate position P5 is regarded as a highly accurate vehicle position. Thereafter, the control unit 15 regards the high-accuracy host vehicle position as the host vehicle position and performs various controls (for example, route guidance and automatic driving control).

  As described above, the in-vehicle device 1 according to the present embodiment detects point cloud data indicating the shape of the feature existing around the vehicle based on the output of the rider 30. The in-vehicle device 1 refers to the feature DB 21 and generates polygon data indicating the shape of the feature recognized from the approximate vehicle position based on the output of the GPS receiver 32 or the like. And the vehicle equipment 1 estimates a highly accurate own vehicle position based on the above-mentioned point cloud data and polygon data. As a result, the in-vehicle device 1 can preferably estimate an accurate vehicle position from the approximate vehicle position recognized based on the outputs of the GPS receiver 32 and the self-supporting positioning sensor 33.

[Modification]
Next, a modified example suitable for the embodiment will be described. The following modifications may be applied in any combination to the above-described embodiments.

(Modification 1)
The control unit 15 may limit the features to be collated in step S105 of FIG. 4 or step S203 of FIG. 5 based on the priorities determined in advance for each feature.

  FIG. 10 shows an example of the data structure of the feature DB 21 referred to in this modification. In the feature DB 21 of FIG. 10, an item “priority” is provided. The “priority” is a priority that is referred to when determining a feature to be collated in the vehicle position estimation process. For example, the priority is determined so that the target feature is less likely to change or disappear (removal), the lidder 30 is less likely to be shielded by the pulse laser, and the map data to be changed or disappeared. It is set in advance in consideration of the ease of reflection. Preferably, features having a height higher than a predetermined height (for example, high-rise buildings) and public facilities such as post offices or public offices such as police stations and fire stations, schools, libraries, hospitals, post offices, etc. As a public object, it may be set to a higher priority than other features.

  When the control unit 15 determines in step S103 in FIG. 4 that there are a plurality of features within the measurement range of the rider 30, the feature for extracting polygon data based on the feature DB 21 based on the above-described priority. Limit. For example, in this case, the control unit 15 may extract only the polygon data corresponding to the feature with the highest priority from the feature DB 21, and only the polygon data corresponding to the feature with the priority higher than a predetermined value. May be extracted from the feature DB 21. In this case, the control unit 15 calculates the similarity indicating the degree to which the polygon data matches a part of the point cloud data in the collation process in step S105 or step S203, and the polygon data is a part of the point cloud data. Is set as a high-accuracy own vehicle position.

  As described above, in the present modification, the control unit 15 determines the feature that is the target to be collated in step S105 in FIG. 4 or step S203 in FIG. 5 based on the priority for each feature. Limited to things. Thereby, the processing load can be reduced and the accuracy of the collation process can be increased.

  As shown in FIG. 10, instead of including the priority information in the feature DB 21, a table in which the priority for each feature type is recorded may be stored in the storage unit 12. In this case, the control unit 15 determines the priority of each feature by referring to the above-described table from the feature type specified by referring to the “feature type” in the feature DB 21.

(Modification 2)
The processing of the flowchart of FIG. 4 may be executed by a control unit (ECU: Electronic Control Unit) (not shown) of the vehicle instead of the in-vehicle device 1. In this case, the control unit of the vehicle electrically connects to various sensors including the rider 30, and executes the flowchart of FIG. 4 by referring to the feature DB 21 stored in a predetermined storage unit.

(Modification 3)
The in-vehicle device 1 collates the three-dimensional data generated based on the output of the camera 31 with the polygon data instead of collating the point cloud data output from the lidar 30 with the polygon data based on the feature DB 21, thereby May be estimated.

  In this case, for example, the camera 31 is a three-dimensional measurement camera having a plurality of image sensors, and the control unit 15 generates three-dimensional data based on the output of the camera 31 in step S101 of FIG. In step S104, the control unit 15 unifies the coordinate system of the polygon data based on the above-described three-dimensional data and the feature DB 21, and collates them in step S105. Similarly, in step S202 and step S203 of FIG. 5, the control unit 15 unifies the coordinate system of the polygon data based on the above-described three-dimensional data and the feature DB 21 based on each candidate position, and in step S203, the matching processing is performed. I do. As described above, the in-vehicle device 1 can perform the vehicle position estimation process according to the present embodiment even when the on-vehicle device 1 is intended for the three-dimensional data obtained from the output of the external sensors other than the rider 30.

DESCRIPTION OF SYMBOLS 1 Onboard equipment 11 Communication part 12 Memory | storage part 13 Sensor part 14 Input part 15 Control part 16 Output part 20 Map data 21 Feature DB

Claims (11)

Detecting means for detecting a first shape which is a shape of a feature existing around the vehicle;
Obtaining means for obtaining a rough vehicle position of the vehicle;
Generating means for generating a second shape which is a shape of the feature recognized from the approximate vehicle position with reference to a feature database storing the position and shape of the feature;
A vehicle position estimation apparatus comprising: estimation means for estimating a highly accurate vehicle position of the vehicle based on the first shape and the second shape. The acquisition means acquires the direction of the vehicle,
The detection means detects the shape of a feature existing in the direction from the vehicle as the first shape,
2. The generation unit generates the shape of a feature existing in the direction as the second shape from the approximate vehicle position or candidate positions set around the approximate vehicle position. The vehicle position estimation device described in 1. The estimation means estimates the approximate vehicle position as the high-accuracy vehicle position when the first shape and the second shape match,
If the first shape and the second shape do not match, the shape of the feature recognized from the candidate positions set around the approximate vehicle position is set as the second shape, and the second shape is the first shape. 3. The vehicle position estimation apparatus according to claim 1, wherein a candidate position that matches the shape is estimated as a highly accurate vehicle position. The generating means generates a second shape for a plurality of candidate positions,
The estimation unit determines a second shape having a maximum similarity to the first shape and a predetermined threshold value or more among a plurality of second shapes as a second shape that matches the first shape. The vehicle position estimation device described in 1.   The generating means is a feature having a predetermined high priority among features recognized from candidate positions of the high-accuracy own vehicle position set in the vicinity of the approximate own vehicle position or the approximate own vehicle position. The vehicle position estimation device according to any one of claims 1 to 4, wherein the vehicle shape is generated as the second shape.   6. The own vehicle position estimation device according to claim 5, wherein the high-priority features include at least one of a feature having a height equal to or higher than a predetermined value and a public property.   The said detection means produces | generates the point cloud data of a feature with an external sensor, and makes the shape specified based on the said point cloud data into the said 1st shape, The own vehicle as described in any one of Claims 1 thru | or 6 Position estimation device.   The own vehicle position estimation apparatus according to any one of claims 1 to 7, wherein the acquisition unit is acquired by a GPS and / or a self-supporting sensor. A control method executed by the vehicle position estimation device,
A detection step of detecting a first shape which is a shape of a feature existing around the vehicle;
An obtaining step of obtaining an approximate vehicle position of the vehicle;
Referring to the feature database storing the position and shape of the feature, a second shape that is a shape of the feature recognized from the approximate own vehicle position or a candidate position that is a peripheral position of the approximate own vehicle position is generated. Generation process;
An estimation step of estimating a high-accuracy host vehicle position of the vehicle based on the first shape and the second shape;
A control method. A program executed by a computer,
Detecting means for detecting a first shape which is a shape of a feature existing around the vehicle;
Obtaining means for obtaining a rough vehicle position of the vehicle;
Referring to the feature database storing the position and shape of the feature, a second shape that is a shape of the feature recognized from the approximate own vehicle position or a candidate position that is a peripheral position of the approximate own vehicle position is generated. Generating means;
A program that causes the computer to function as estimation means for estimating a highly accurate vehicle position of the vehicle based on the first shape and the second shape.   A storage medium storing the program according to claim 10.

Images