JP7284851B2 - Information processing equipment - Google Patents

Description

The present invention relates to an information processing device.

In recent years, vehicles (self-driving vehicles) have been developed that run independently while acquiring the surrounding conditions of the vehicle using various sensors such as cameras and lidars (LiDAR: Light Detection And Ranging) mounted on the vehicle.

Map data with accurate road information is required for this type of self-driving vehicle to run in self-driving mode (see Patent Document 1, for example). Accurate road information includes, for example, not only detailed positional information such as road width, lanes, and signs, but also information such as road slope, road unevenness, and road shoulder unevenness. Generated as map data.

JP 2016-43747 A

The three-dimensional map for automatic driving mentioned above is expensive to create. In addition, since such a three-dimensional map for automatic driving is generated using, for example, point group information obtained by a lidar or the like, the amount of data increases and a large-capacity storage device or the like is required. Therefore, automatic driving using a two-dimensional map based on point cloud information and the like is under consideration.

When a two-dimensional map is used for automatic driving, since this two-dimensional map is raster information generated by projecting point group information, there is no height direction information. Therefore, for example, when traveling on a position where roads with different heights overlap, such as overpasses and underpasses, there is a risk that the autonomous vehicle will not be able to correctly recognize its position.

One example of the problem to be solved by the present invention is to identify roads that overlap in the height direction in a two-dimensional map as described above.

An information processing apparatus according to the present invention is map data used by an autonomous vehicle during autonomous driving, and includes a plurality of two-dimensional map data divided into areas having a certain area, and includes a first road and a second road. A storage unit that stores map data in which the first road and the second road are represented by different two-dimensional map data for the area where the road overlaps in the height direction, and the autonomous driving vehicle based on the map data. and a control unit for causing the vehicle to travel, wherein the control unit refers to the two-dimensional map data showing the road on which the vehicle is currently traveling, out of the first road and the second road. roads and drive self-driving vehicles autonomously.

1 is a schematic configuration diagram of a system having a map data storage device according to a first embodiment of the present invention; FIG. 2 is a functional configuration diagram of the server device shown in FIG. 1; FIG. FIG. 3 is a flow chart of an operation for generating two-dimensional map data in the server device shown in FIG. 2; FIG. FIG. 4 is an explanatory diagram of association between links and travel loci; FIG. 4 is an explanatory diagram of intersecting links; FIG. 4 is an explanatory diagram of generation of two-dimensional map data; FIG. 4 is a layout image diagram of two-dimensional map data with different height directions at the same spot. FIG. 9 is an explanatory diagram of links that are targets of the map data generating method according to the second embodiment of the present invention; 9 is an explanatory diagram of an example of division of the link shown in FIG. 8; FIG. FIG. 10 is a flow chart of the operation of generating two-dimensional map data according to the second embodiment of the present invention; FIG. 11 is an example of two-dimensional map data generated by executing the flowchart shown in FIG. 10; 4 is a flow chart until the automatic driving vehicle starts running when the automatic driving is performed.

A map data structure according to an embodiment of the present invention will be described below. A map data structure according to an embodiment of the present invention includes a plurality of two-dimensional map data divided for each area, and for an area where a first road and a second road overlap in a height direction, the first road and the second road Two roads are represented by different two-dimensional map data. By doing so, the two-dimensional map data for the area where the first road and the second road overlap in the height direction are different. By referring to it, it is possible to identify roads that overlap in the height direction.

In addition, the plurality of two-dimensional map data may be divided into areas for each predetermined section of the road. For example, by having two-dimensional map data for each link, it is possible to travel even in areas overlapping other links. It is possible to identify the road (link) that is

In addition, the plurality of two-dimensional map data may be divided into areas having a certain area, and the map data can be configured by arranging the areas having a certain area, for example, in tiles. .

In addition, since each of the plurality of two-dimensional map data has identification information indicating relative height, for example, a spiral road can be divided into a plurality of sections and have identification information for each section. , it is possible to correctly recognize its own position.

A map data storage device according to an embodiment of the present invention stores the map data structure described above. By doing so, it is possible to identify roads that overlap in the height direction during automatic driving by storing the map data in a server device or the like that distributes the map data to the automatic driving vehicle or the like.

Further, in the map data generation method according to the embodiment of the present invention, in the traveling locus acquisition step, the traveling locus traveled by the vehicle that acquires the recognition information that is the recognition result of the feature is acquired, and in the associating step, the traveling locus is obtained. Associate with the section that the road network information has. Next, in the recognition information acquisition step, recognition information is acquired for each section associated in the association step, and two-dimensional map data is generated in the generation step based on the recognition information acquired in the recognition information acquisition step. By doing so, it is possible to associate the travel locus when acquiring the recognition information necessary for generating the two-dimensional map data such as the point cloud information with the link of the existing road network information. Therefore, since two-dimensional map data can be generated for each section such as a link, it is possible to identify individual roads even if the roads overlap in the height direction. It is also possible to use information such as altitude that links have. Furthermore, by using existing road network information, it is possible to reduce the cost of generating map data for automatic driving.

Also, the generating step may generate two-dimensional map data for each link in the road network information. By doing so, two-dimensional map data is generated for each link, so even roads that intersect can be identified if they have different links.

The road network information may further include a division step of dividing the section into a plurality of sections, and the association step may associate each section divided by the division step. By doing so, for example, when there is an intersecting part in one link such as a spiral road, the relative height position can be correctly recognized by dividing the link into a plurality of parts. becomes possible.

Further, in the dividing step, identification information in the height direction may be assigned to each of the divided sections. By doing so, for example, by dividing a spiral road into a plurality of sections and having identification information for each section, it is possible to correctly recognize the own position.

A map data structure and a feature information generating method according to a first embodiment of the present invention will be described with reference to FIGS. 1 to 6. FIG. A server device 1 as a feature data storage device according to the present embodiment can communicate with an information processing device 3 mounted on a measurement vehicle C via a network N such as the Internet, as shown in FIG. ing.

The measurement vehicle C has an information processing device 3 and a sensor 4, and is a vehicle that measures (obtains) recognition information such as the surrounding environment with the sensor 4 by traveling on a predetermined route. , self-driving vehicles.

The information processing device 3 uploads information based on the surrounding information detected by the sensor 4 to the server device 1 . The information processing device 3 may be configured as part of an in-vehicle device such as a navigation device, or may be configured as a single device. Alternatively, a portable terminal device such as a notebook PC may be used as long as the sensor 4 can be connected.

The sensor 4 is a sensor for recognizing the information of the vehicle such as the position of the vehicle and the surrounding environment (features existing in the vicinity), and includes a camera, a lidar, a GPS (Global Positioning System) receiver, an acceleration Including sensors, speed sensors and the like. A camera produces a colored image that represents the situation in the outside world. A lidar discretely measures the distance to an object existing in the outside world and recognizes the position, shape, etc. of the object as a three-dimensional point group. The GPS receiver produces latitude and longitude location information representing the current vehicle location. The acceleration sensor detects acceleration of the vehicle. A speed sensor detects the speed of the vehicle. Note that the sensor 4 may include an inertial measurement unit (IMU), a gyro sensor, or the like for recognizing the attitude (orientation, etc.) of the vehicle and correcting data acquired by other sensors. Furthermore, the sensor 4 may include a raindrop sensor, a thermometer, a hygrometer, etc. for detecting whether it is raining. Information about the vehicle such as the position of the vehicle and information about the surrounding environment detected and acquired by the sensor 4 is output to the information processing device 3 as peripheral information.

The sensor 4 can obtain the feature recognition results necessary for generating map data. A feature is a concept that includes any natural or man-made object that exists on the ground. Examples of features include on-route features located on the route of the vehicle (that is, the road) and surrounding features located around the road. Examples of on-route features include road signs, traffic lights, guardrails, footbridges, and the like, including roads themselves. That is, characters and figures drawn on the road surface, and the shape of the road (road width and curvature) are also included in the on-route features. Examples of surrounding features include buildings (houses, shops) and signboards located along roads.

FIG. 2 shows the functional configuration of the server device 1 as a map data storage device. The server device 1 includes a control section 11 , a communication section 12 and a storage section 13 .

The control unit 11 functions as a CPU (Central Processing Unit) of the server device 1 and controls the entire server device 1 . The control unit 11 stores the information uploaded from the information processing device 3 in the storage unit 13 as acquired information 13a, and generates two-dimensional map data 13c based on the acquired information 13a and road network information 13b, which will be described later. .

The communication unit 12 functions as a network interface or the like of the server device 1 and receives information uploaded by the information processing device 3 . In this embodiment, the information measured by the sensor 4 is uploaded by the information processing device 3 through communication. good too.

The storage unit 13 functions as a storage device such as a hard disk of the server device 1, and stores the obtained information 13a, the road network information 13b, and the two-dimensional map data 13c. The road network information 13b includes information on a network represented by a combination of nodes, which are nodes on a transportation network representation, and links connecting the nodes. This road network information 13b may be, for example, map data used in a navigation device or the like for guiding a route to a destination.

The two-dimensional map data 13c is map data generated by the control unit 11 based on the obtained information 13a and road network information 13b described later. The two-dimensional map data 13c in this embodiment is raster information generated by projecting point group information uploaded from the information processing device 3, for example. In this embodiment, a two-dimensional map mainly used for automatic driving will be described, but the present invention is not limited to this, and may be a two-dimensional map for use in a navigation device for a manually operated vehicle or various map services.

In the configurations shown in FIGS. 1 and 2, the server device 1 acquires the acquired information 13a, has the road network information 13b, and generates the two-dimensional map data 13c. may be distributed to other server devices or the like.

Next, the operation of generating two-dimensional map data (map data generating method) in the server apparatus 1 having the above configuration will be described with reference to the flowchart of FIG. The flowchart of FIG. 3 is executed by the control unit 11 of the server device 1 . That is, the server device 1 becomes a map data generation device.

First, in step S1, the measurement vehicle C is driven and the sensor 4 measures (obtains) information based on the surrounding information of the measurement vehicle C (recognition information, which is the recognition result of the feature, and the travel locus). The measurement results are uploaded from the information processing device 3 to the server device 1 and accumulated in the storage unit 13 as acquired information 13a. In this embodiment, the acquired information 13a includes the point cloud information and the travel locus of the measurement vehicle C. FIG. The point group information is information on a three-dimensional point group recognized by the lidar included in the sensor 4 . The travel locus is information generated from the position information of the vehicle C measured by the GPS receiver included in the sensor 4, and includes, for example, latitude/longitude information, altitude information, and time information.

Next, in step S2, a matching result d1 is generated by matching the travel locus of the measured vehicle C with links as sections possessed by the road network (NW) information 13b. A matching method will be described with reference to FIG. In FIG. 4, there are three nodes, nodes N1 to N3, as the road network information 13b. Link B is between the nodes N1 and N2, and link A is between the nodes N2 and N3. Then, it is assumed that the travel locus (acquired information 13a) of the measurement vehicle C traveling in the order of node N1, link B, node N2, link A, and node N3 is obtained (see the dashed line in FIG. 4).

In this case, the position information of the node N1 and the position information included in the travel locus are compared, and if they match or are within a predetermined range, it is determined that the node N1 has been passed. After that, the position information of the node N2 and the position information included in the travel locus are collated, and if they match or are within a predetermined range, it is determined that the node N2 has been passed. Therefore, it is determined that the vehicle traveled on the link B from the time included in the travel locus determined to have traveled through the node N1 to the time included in the travel locus determined to have traveled through the node N2. Similarly, it is determined that the vehicle traveled on the link A from the time included in the traveling locus determined to have passed the node N2 to the time included in the traveling locus determined to have passed the node N3. In addition, when the link has position information, the position information of the link and the travel locus may be collated. In this way, the links included in the road network information 13b are associated with the travel locus, and the result of this association is the matching result d1.

Next, in step S3, two-dimensional map data 13c is generated based on the matching result d1 generated in step S2 and the point group information (acquired information 13a). Two-dimensional map data is generated for each matched (associated) link. For example, in the case of link A, the point cloud information acquired from the time included in the traveling locus determined to have passed node N2 to the time included in the traveling locus determined to have passed node N3 is read out. , the point group information is projected by a well-known method to generate two-dimensional map data. By doing so, the two-dimensional map data can be generated with only the point group information of the portion corresponding to the link A. FIG. Therefore, for example, as shown in FIG. 5, if there is a link C that intersects (overlaps) with link B but cannot be directly moved between the links, link B and link C are two different two-dimensional map data. Since it can be used as dimensional map data (file), the running position (height position) can be recognized without error at the point where the link B and the link C intersect.

Also, the two-dimensional map data may be generated by dividing it into a plurality of areas in the horizontal direction other than the height direction described above. The dividing method may be dividing for each node, or, as shown in FIG. 6, dividing into rectangles (tiles) having a predetermined length in the latitude and longitude directions, for example. That is, it may be divided into regions having a constant area. FIG. 6 is an example in which FIG. 5 is divided into four regions. In the example of FIG. 6, the link A spans two areas, one for the file f1 and the other for the file f2. Link B spans two areas, one for file f3 and one for file f4. The link C spans two areas, one for file f5 and one for file f6. These areas are generated as files f1 to f6 as different two-dimensional map data.

For example, a file f1 containing the node N2 of the link A and a file f3 containing the node N2 of the link B are two-dimensional map data indicating the same area, but are different files. File f4 including node N1 of link B and file f5 including link C are two-dimensional map data indicating the same area, but are different files.

FIG. 7 is an image diagram in which files f4 and f5, which are two-dimensional map data showing the same area, are superimposed in the height direction. FIG. 7 shows a case where link C intersects link B as an actual road or the like. In this case, the autonomous vehicle will refer to different files when traveling on link B and when traveling on link C. can be identified. That is, for an area where link B (first road) and link C (second road) overlap in the height direction, link B (first road) and link C (second road) are displayed using different two-dimensional map data. is represented.

The two-dimensional map data 13c shown in FIG. 2 has a plurality of files (two-dimensional map data) such as files f1 to f6 described above, and has a map data structure according to this embodiment. The map data structure may simply have a plurality of files of two-dimensional map data, or may have a layer structure consisting of a plurality of files as shown in FIG.

As is clear from the above description, step S1 is the travel locus acquisition process, step S2 is the association process, and step S3 is the generation process.

According to this embodiment, a plurality of two-dimensional map data (files) divided for each area is included. is represented by By doing so, since different files are used for the area where the link B and the link C overlap in the height direction, when driving, by referring to the file corresponding to each road, the height direction can be changed. It is possible to correctly recognize overlapping roads.

In addition, the plurality of files may be divided into areas for each predetermined link on the road, and it is possible to prevent erroneous recognition of the road (link) on which the vehicle is traveling even in areas where other links overlap. Also, the plurality of files may be divided for each area having a certain area, and by arranging the areas having a certain area in a tile form, the map data can be configured.

The server device 1 also stores the two-dimensional map data 13c described above. By doing so, by storing the map data in order to distribute the map data to the automatically driving vehicle or the like, it is possible to correctly recognize the roads that overlap in the height direction during automatic driving.

Further, in step S1, the traveling locus of the vehicle that obtains the recognition information that is the recognition result of the feature is acquired, and the traveling locus obtained as a result of the measurement in step S1 is matched with the links included in the road network information 13b. and associate. Next, in step S3, point group information is acquired based on the result of matching in step S2, and two-dimensional map data 13c is generated based on the point group information. By doing so, it is possible to associate the travel locus when acquiring the point group information with the existing road network information 13b. Therefore, since the two-dimensional map data 13c can be generated for each link, individual roads can be identified even if the roads overlap in the height direction. It is also possible to use information such as altitude that links have. Furthermore, by using existing road network information, it is possible to reduce the cost of generating map data for automatic driving.

Also, in step S3, the two-dimensional map data 13c is generated for each link in the road network information 13b. By doing so, it is possible to generate two-dimensional map data for each link of the road network information 13b.

Next, a map data structure and map data generation method according to a second embodiment of the present invention will be described with reference to FIGS. 8 to 11. FIG. The same reference numerals are given to the same parts as those of the first embodiment described above, and the description thereof will be omitted.

In the first embodiment, the matching process is performed for each link. However, in the case of a spiral link as shown in FIG. 8, for example, an intersecting point (self-intersection) occurs within one link. Therefore, matching processing cannot be performed for each link as in the first embodiment. This embodiment is an embodiment in which the first road and the second road are the same link. That is, the present invention can be applied even if the first road and the second road are the same link.

Therefore, in this embodiment, one link is subdivided into a plurality of sections at points where self-intersections occur. In the example of FIG. 8, there are two self-intersection points c1 and c2, so the line is divided into four sections as shown in FIG. At this time, the divided section is indicated by a ratio to the length of the original link. In the case of FIG. 9, the interval shown at the bottom is 0 or more and less than 0.3. This indicates the length from one end of the original link to the 30% position. Similarly, the second from the bottom is 0.3 or more and less than 0.5 (the length from the 30% position to the 50% position from one end), and the third from the bottom is 0.5 or more and 0.8 Less than, the highest is 0.8 to 1. In the example of FIG. 9, numbers 0 to 1 are used to divide one link into a plurality of sections, but the notation method for subdivision is not limited to this. Other numbers, letters, etc. may be used as long as the ratio of the length of each link into which one link is divided into a plurality of sections can be known. Also, the positional coordinates of the start point and the end point of each divided link may be used, or the length or the like calculated from the positional coordinates of the start point and the end point of each divided link may be used.

A method of generating two-dimensional map data (map data generating method) according to the present embodiment will be described with reference to the flowchart of FIG. In the flowchart of FIG. 10, steps S1 and S2, the acquired information 13a, and the road network information 13b are the same as in the flowchart shown in FIG.

In step S11 after acquiring the travel locus (acquired information 13a), it is determined whether or not there is a link having a self-intersection as described above. Proceed to step S12, and if there is no link having a self-intersection (NO in S11), proceed to step S2. Whether or not there is a self-intersection may be determined based on the position information, altitude information, etc. of the link included in the road network information 13b.

At step S12, the link is divided into a plurality of sections by the method shown in FIGS. For the divided sections, step S2 described in FIG. 3 is executed to perform matching processing for each section. For each section, as shown in FIG. 9, the ratio to the link length is known, so matching with the travel locus is possible. That is, step S12 is a dividing step of further dividing the link (section) into a plurality of sections.

Then, when generating two-dimensional map data from the matching result d1 after the matching process (step S3A), as shown in FIG. Add relative height information as information. In FIG. 11, "0" is assigned to file f10, "1" is assigned to file f11, "2" is assigned to file f12, and "3" is assigned to file f13. In FIG. 11, the relative height information is continuous integers (numerical values), but other identifiable information such as alphabets may be used. With this relative height information, it is possible to easily acquire the next two-dimensional map data when traveling on this link. Also, the relative height can be easily determined. That is, identification information in the height direction is assigned to each of the divided sections.

According to this embodiment, the plurality of two-dimensional map data (files f10 to f13) each have relative height information indicating height. By dividing and having relative height information for each, it becomes possible to correctly recognize its own position.

Further, a step S12 of dividing the link in the road network information 13b into a plurality of sections is further included, and the step S2 associates each section divided by the step S12. By doing so, for example, when there is an intersecting part in one link such as a spiral road, the relative height position can be correctly recognized by dividing the link into a plurality of parts. becomes possible.

Finally, the use scene of the two-dimensional map data 13c (map data structure) described above will be described with reference to FIG. FIG. 12 is a flowchart until the automatic driving vehicle starts running when performing automatic driving. First, a route is planned using a terminal device or the like mounted on an automatic driving vehicle (step S21). That is, a start point (eg, current location) and an end point (destination) are determined. A route search is performed using the determined start point and end point as data d2 (step S22). A route search may be performed by a well-known route search algorithm. A route to be traveled is determined from the route candidate d3 extracted as a result of the route search (step S23). The determination method may be selected by a user or the like, or may be determined automatically based on traffic conditions such as congestion.

Next, necessary two-dimensional map data is acquired from the two-dimensional map data 13c based on the determined route d4 (step S24). Note that the two-dimensional map data 13c may be not the data acquired from the storage unit 13 of the server device 1, but the two-dimensional map data stored in a terminal device or the like mounted on the automatic driving vehicle. Then, traveling is started using the acquired two-dimensional map data d5 (step S25).

In the above-described embodiment, the position where the roads intersect has been described, but the position is not limited to this as long as the first road and the second road overlap. For example, by applying the present invention, it is possible to correctly recognize the road on which the vehicle is traveling, even if there is an expressway on top of the general road.

Moreover, the present invention is not limited to the above embodiments. That is, those skilled in the art can carry out various modifications according to conventionally known knowledge without departing from the gist of the present invention. As long as the map data structure and the map data generation method of the present invention are provided, such modifications are of course included in the scope of the present invention.

1 server device (map data storage device, map data generation device)
13c Two-dimensional map data (map data structure)
S1 measurement (running locus acquisition step)
S2 Matching to link (association process)
S3 Map data generation (recognition information acquisition process, generation process)
S12 division (dividing step)

Claims (2)

Map data used by an autonomous vehicle during autonomous driving, including a plurality of two-dimensional map data divided into areas having a certain area, where the first road and the second road overlap in the height direction. a storage unit that stores map data in which the first road and the second road are represented by different two-dimensional map data;
An information processing device comprising a control unit that causes the automatically driven vehicle to autonomously travel based on the map data,
The plurality of two-dimensional map data includes a plurality of two-dimensional map data generated for each of a plurality of links that intersect but cannot be directly accessed between the links;
The control unit identifies the currently traveling road by referring to the two-dimensional map data representing the currently traveling road, out of the first road or the second road, and identifies the road on which the automated driving vehicle is traveling. to run autonomously,
An information processing device characterized by: 2. The information processing apparatus according to claim 1, wherein each of said plurality of two-dimensional map data has identification information indicating relative height.

Images