Information processing device

Abstract

To generate a road graph based on probe data collected by a mobile body.SOLUTION: An information processing device performs: acquiring first map data including position information of a road; acquiring probe data including aggregation of position information of a first mobile body positioned by the first mobile body moving on the road; making machine learning model trained with the first map data as teacher data and with the probe data as input data; and converting second probe data including aggregation of position information of a second mobile body into a road graph including the position information of the road with the use of the leaned machine learning model.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present disclosure relates to the collection of road information.

Background Art

[0002] There is a technique of using probe data collected by vehicles to generate highly accurate map data. In this regard, for example, Patent Document 1 discloses a system for updating an existing digital road network based on accumulated probe data.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Patent Document 2

Patent Document 3

Summary of the Invention

Problems to be Solved by the Invention

[0004] An object of the present disclosure is to generate a road graph based on probe data collected by a moving object.

Means for Solving the Problems

[0005] One aspect of an embodiment of the present disclosure is Obtaining first map data including road location information, obtaining probe data including a set of location information of the first moving body measured by the first moving body moving on the road, training a machine learning model using the first map data as teacher data and the probe data as input data, and converting second probe data including a set of location information of a second moving body into a road graph including road location information using the trained machine learning model, an information processing apparatus having a control unit that executes the above.

[0006] Also, as another aspect, there is a method executed by the above information processing apparatus, a program for causing a computer to execute the method, or a computer-readable storage medium storing the program non-temporarily.

Advantages of the Invention

[0007] According to the present disclosure, a road graph can be generated based on probe data collected by a moving body.

Brief Description of the Drawings

[0008]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Mode for Carrying Out the Invention

[0009] In order to generate highly accurate road map data, there is an attempt to utilize probe data collected by vehicles. For example, position information can be periodically collected from a plurality of vehicles, and based on the collected position information, the locations where roads exist can be estimated. According to such a configuration, for example, when there is a newly opened road, vehicle manufacturers can quickly grasp that the number of drivable road links has increased without waiting for data updates by map-providing companies.

[0010] On the other hand, when utilizing probe data collected by vehicles, the accuracy of the data becomes a problem. For example, since the GPS module installed in a vehicle has an error of about several meters, if the data is used as it is, it may not be possible to correctly determine the road area in some cases.

[0011] Here, an example is given. FIG. 6(A) is a diagram showing a plurality of position information obtained by a vehicle traveling on a certain road. As shown in the figure, the position information obtained by the vehicle has a large error and may deviate from the road. Therefore, even if they are connected (shown by a dotted line), it is not always possible to correctly represent the road shape. Therefore, when generating a road map based on such data, an area that is not a road may be represented as a road, or an area where a road exists may be represented as outside the road in some cases. The information processing apparatus according to the present disclosure solves such problems.

[0012] An information processing apparatus according to an embodiment is Obtaining first map data including road location information, obtaining probe data including a set of position information of the first moving object measured by the first moving object moving on the road, training a machine learning model using the first map data as teacher data and the probe data as input data, and converting second probe data including a set of position information of a second moving object into a road graph including road location information using the trained machine learning model, and having a control unit that executes the above.

[0013] The first map data is data including a road map that serves as teacher data for training a machine learning model. The first map data can typically be a representation of a road as a graph (road graph). A road graph is a geospatial graph in which edges represent roads and nodes represent intersections. The road graph may include road location information. For example, the first map data may be data representing a road network based on the position of the center line of the road.

[0014] The probe data is data including a set of position information of the first moving object (e.g., a probe car). The position information may be sensed by the probe car.

[0015] The control unit trains a machine learning model using the first map data as teacher data and the probe data as input data. As a result, a machine learning model that has learned the relative positional relationship between the position information of the probe car and the road (e.g., the center line of the road) can be obtained. The machine learning model is, for example, a model that outputs a set of points corresponding to the center of a road when a set of position information of a vehicle traveling on a certain road is input.

[0016] The control unit converts second probe data including a set of position information of a second moving object into a road graph using the trained machine learning model. For example, based on probe data collected from a probe car that has passed through the road shown in Fig. 6(A), it is assumed that the positional relationship between the position information and the road is learned. When the position information of a vehicle traveling on the same road is input into the machine learning model thus obtained, as shown in Fig. 6(B), a set of points corresponding to the center of the estimated road can be obtained. As shown, a set of points corresponding to the center of the estimated road can be obtained.

[0017] Furthermore, the machine learning model can learn the relative positional relationship between a plurality of pieces of position information and the actual road. Thereby, when a vehicle travels along a certain trajectory, the shape of the actual road can be estimated. Thereby, for example, even when the second probe data is generated in an area where the first moving body has not traveled (an area not included in the teacher data), if a similar driving trajectory has been learned, the actual road shape can be estimated.

[0018] Note that the road graph may be represented by the center line of the road. Also, the control unit may cause the machine learning model to learn using the center line of the road included in the first map data as teacher data.

[0019] Also, the probe data and the second probe data may further include data regarding the positions of road boundaries and / or lane boundaries. By adding, to the input data, in addition to the position information of the vehicle, a line indicating the boundary of the road (road boundary line) and a line indicating the boundary of the lane (lane boundary line), the accuracy of the estimation can be improved. The positions of the road boundary line and the lane boundary line may be those detected by the moving body (for example, those obtained by analyzing, on the vehicle side, an image acquired by an in-vehicle camera).

[0020] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. The configurations of the following embodiments are examples, and the present disclosure is not limited to the configurations of the embodiments.

[0021] (First Embodiment) The outline of the vehicle system according to the first embodiment will be described. The vehicle system according to this embodiment includes a server device 10 that generates a road graph, a vehicle (vehicle 1) that supplies probe data to the server device 10, and a server (map server 20) that supplies road map data for learning to the server device 10.

[0022] The server device 10 generates a machine learning model based on the road map data supplied from the map server 20, and uses the machine learning model to generate a road graph including the position information of the road. The road graph generated by the server device 10 may be outside the range of the road map data for learning.

[0023] Referring to FIG. 1, the outline of the processing performed by the system will be described. The processing performed by the system can be divided into a learning phase and an estimation phase. FIG. 1(A) is a diagram for explaining the learning phase. The server device 10 acquires road map data for performing machine learning from the map server 20. The road map data provided by the map server 20 represents a road network included in a predetermined area by a geospatial graph (nodes and edges). The road map data can be, for example, data representing the position of the center line of the road. In this embodiment, the road map data provided by the map server 20 is referred to as a master map.

[0024] The server device 10 is configured to be communicable with a plurality of vehicles 1, and acquires a plurality of position information collected by each vehicle from the plurality of vehicles 1. In this embodiment, the plurality of vehicles 1 periodically collect position information and transmit it to the server device 10. The plurality of position information can also be said to be information representing the driving trajectory of the vehicle 1.

[0025] Further, the server device 10 performs machine learning using the set of the collected position information as input data and the position of the center line of the road included in the master map as teacher data, and generates a machine learning model. When the set of position information of the vehicle 1 is input, the machine learning model estimates the position of the center line of the road on which the vehicle 1 travels and outputs a set of corresponding points. The machine learning model is referred to as a "road model".

[0026] FIG. 1(B) is a diagram for explaining the estimation phase. Similar to the learning phase, the server device 10 acquires a plurality of pieces of position information collected by each vehicle from a plurality of vehicles 1. The plurality of vehicles 1 do not necessarily have to be the vehicles that acquired the position information in the learning phase. The server device 10 inputs the set of position information into the road model of the learned machine learning model, estimates the position of the center line of the road on which the vehicle 1 travels based on the output from the road model, and generates a road graph including the estimated center line position.

[0027] [Device Configuration] FIG. 2 is a diagram showing an example of the configurations of the server device 10, the map server 20, and the vehicle 1. First, the server device 10 will be described. The server device 10 is a computer such as, for example, a personal computer, a smartphone, a mobile phone, a tablet computer, or a personal information terminal. The server device 10 includes a control unit 11, a storage unit 12, a communication unit 13, and an input / output unit 14.

[0028] The server device 10 can be configured as a computer having a processor (CPU, GPU, etc.), a main storage device (RAM, ROM, etc.), and an auxiliary storage device (EPROM, hard disk drive, removable media, etc.). The auxiliary storage device stores an operating system (OS), various programs, various tables, etc., and by executing the programs stored therein, various functions (software modules) that meet predetermined purposes, as described later, can be realized. However, some or all of the functions may be realized as hardware modules by hardware circuits such as ASICs, FPGAs, etc.

[0029] The control unit 11 is an arithmetic unit that realizes various functions of the server device 10 by executing a predetermined program. The control unit 11 can be realized by a hardware processor such as a CPU, for example. Also, the control unit 11 may be configured to include a RAM, a ROM (Read Only Memory), a cache memory, and the like.

[0030] The control unit 11 is configured to have three software modules: an acquisition unit 111, a learning unit 112, and an estimation unit 113. Each software module may be realized by executing a program stored in the storage unit 12 described later by the control unit 11 (such as a CPU).

[0031] The acquisition unit 111 periodically acquires vehicle data from a plurality of vehicles 1 under its management. The vehicle data is data related to driving generated by the vehicle 1 and includes the position information of the vehicle 1. The acquisition unit 111 periodically acquires vehicle data from the plurality of vehicles 1 and stores it in the storage unit 12 described later. By referring to the stored vehicle data, the history of the position information of each vehicle (that is, the driving trajectory) can be obtained. Note that since the position information of each vehicle is obtained by a GPS device or the like, it is not always accurate.

[0032] In the present embodiment, two types of vehicle data are exemplified: one for learning a road model and the other for generating a road graph using the learned road model. The former is referred to as first vehicle data, and the latter is referred to as second vehicle data. Also, the vehicle that collects the first vehicle data is referred to as the first vehicle, and the vehicle that collects the second vehicle data is referred to as the second vehicle.

[0033] Also, the acquisition unit 111 receives a master map from the map server 20. The master map is teacher data for learning a road model and is typically graph data representing road edges by the center line of the road. Note that the road edges may be defined for each driving direction. That is, a road that can be traveled in both directions may be represented by two road edges. In this case, there are two center lines.

[0034] The learning unit 112 learns a road model based on the first vehicle data and the master map. FIG. 3(A) is a diagram for explaining the learning of the road model. The learning unit 112 extracts the first vehicle data collected by the first vehicle for learning from the storage unit 12, and obtains the set of position information included in the first vehicle data as input data. Also, from the master map, the area where the first vehicle has traveled is extracted, and the center line of the road is obtained as teacher data. Then, the learning unit 112 executes the learning of the road model using the input data and the teacher data.

[0035] Thereby, when a plurality of pieces of position information acquired by a vehicle traveling on a certain road are input, a machine learning model can be obtained that estimates the position of the center line of the road and outputs a corresponding set of position information. In addition, since the road model learns the relative positional relationship between the position information acquired by GPS and the actual road, position information of an area not included in the master map can also be taken as input data. Also in this case, the result of estimating the center position of the road is output from the road model.

[0036] The estimation unit 113 generates a road graph of an arbitrary area using the learned road model. As described above, the road model learns the relative positional relationship between the position information of the vehicle and the actual road. Therefore, by inputting the set of position information of the vehicle 1 (second vehicle) that has traveled in an arbitrary area, the center position of the road traveled by the second vehicle can be estimated. By executing this for a plurality of second vehicles, a road graph corresponding to the road within the area can be obtained.

[0037] FIG. 3(B) is a diagram for explaining the processing performed by the estimation unit 113. The estimation unit 113 extracts the second vehicle data collected in the area for which the road graph is to be generated from the storage unit 12, and acquires the set of position information included in the second vehicle data as input data. Then, the estimation unit 113 inputs the acquired input data into the road model and obtains the output. The road model outputs a set of points (position information) representing the center of the road.

[0038] Note that since a set of position information is output as points from the road model, it cannot be used as a road graph as it is. Therefore, the estimation unit 113 converts the set of points into a road graph by correction processing described later.

[0039] The storage unit 12 is a means for storing information, and is composed of a storage medium such as a RAM, a magnetic disk, or a flash memory. The storage unit 12 stores programs executed by the control unit 11, data used by the programs, and the like.

[0040] The storage unit 12 stores the first vehicle data, the second vehicle data, the master map, and the road model described above.

[0041] The communication unit 13 is a wireless communication interface for connecting the server device 10 to a network. The communication unit 13 is configured to be able to communicate with the map server 20 and the vehicle 1 via, for example, a mobile communication service such as a wireless LAN, 3G, 4G, or 5G.

[0042] The input / output unit 14 is a unit that receives input operations performed by the operator of the device and presents information to the operator. In the present embodiment, it consists of a single touch panel display. That is, it is composed of a liquid crystal display and its control means, and a touch panel and its control means.

[0043] Note that the specific hardware configuration of the server device 10 can be appropriately omitted, replaced, or added according to the embodiment. For example, the control unit 11 may include a plurality of hardware processors. The hardware processors may be composed of a microprocessor, FPGA, GPU, etc. Also, input / output devices other than those exemplified (e.g., optical drive, etc.) may be added. Further, the server device 10 may be composed of a plurality of computers. In this case, the hardware configurations of the respective computers may or may not match.

[0044] The map server 20 is a device that provides road map data (master map) within a specified range in response to a request from the server device 10. The road map data can be data representing a road network by edges and nodes. In the present embodiment, the edges included in the road map data represent the positions of the center lines of the roads. The map server 20 can also be configured by a computer having a processor and a storage device similar to those of the server device 10.

[0045] The vehicle 1 is equipped with an in-vehicle device 30. The in-vehicle device 30 is a computer that provides a predetermined function to the passengers of the vehicle 1. The in-vehicle device 30 may be, for example, a car navigation device or a head unit, etc. In the present embodiment, the in-vehicle device 30 has a function of periodically generating information (vehicle data) related to the vehicle 1 and transmitting it to the server device 10.

[0046] The in-vehicle device 30 can be configured as a computer having a processor (such as a CPU or GPU), a main storage device (such as a RAM or ROM), and an auxiliary storage device (such as an EPROM, a hard disk drive, or a removable medium). The auxiliary storage device stores an operating system (OS), various programs, various tables, etc. By executing the programs stored therein, various functions (software modules) that meet a predetermined purpose, as described later, can be realized. However, some or all of the functions may be realized as hardware modules by a hardware circuit such as an ASIC or an FPGA.

[0047] The in-vehicle device 30 includes a control unit 31, a storage unit 32, a communication unit 33, and a position information acquisition unit 34.

[0048] The control unit 31 is an arithmetic unit that realizes various functions of the in-vehicle device 30 by executing a predetermined program. The control unit 31 can be realized by a hardware processor such as a CPU, for example. Further, the control unit 31 may include a RAM, a ROM (Read Only Memory), a cache memory, etc.

[0049] The control unit 31 is configured to have a message transmission unit 311 as a software module. The software module may be realized by executing, by the control unit 31 (such as a CPU), a program stored in the storage unit 32 described later.

[0050] The message transmission unit 311 periodically generates vehicle data and transmits it to the server device 10. The vehicle data is data related to the running of the vehicle 1, and includes, for example, the speed of the vehicle 1, the traveling direction, the position information, etc. When the timing for generating the vehicle data arrives, the message transmission unit 311 acquires the position information of the vehicle 1 via the position information acquisition unit 34 described later. Further, it generates vehicle data including the acquired position information and transmits it to the server device 10.

[0051] The storage unit 32 is a means for storing information, and is composed of a storage medium such as a RAM, a magnetic disk, or a flash memory. The storage unit 32 stores a program executed by the control unit 31, data used by the program, and the like.

[0052] The communication unit 33 is a device for performing wireless communication with a predetermined network. In the present embodiment, the communication unit 33 is configured to be connectable to a predetermined cellular communication network. The communication unit 33 is configured to have an eUICC (Embedded Universal Integrated Circuit Card) for identifying a user. The eUICC may be a physical SIM card or an eSIM or the like.

[0053] The position information acquisition unit 34 acquires the position information of the vehicle 1. The position information acquisition unit 34 includes a GPS antenna and a positioning module for positioning the position information. The GPS antenna is an antenna that receives a positioning signal transmitted from a positioning satellite (also referred to as a GNSS satellite). The positioning module is a module that calculates position information based on the signal received by the GPS antenna.

[0054] [Outline of correction processing] Next, an outline of the correction processing performed by the above-described estimation unit 113 will be described. FIG. 4(A) is a schematic diagram of a road network existing in a certain area. In this example, it is assumed that there are three roads respectively indicated by reference numerals 401, 402, and 403. Here, it is assumed that a road model performs estimation based on vehicle data (second vehicle data) collected from a plurality of vehicles that have traveled on the three roads, and outputs the result.

[0055] Figure 4(B) shows the result of plotting the set of position information output by the road model (i.e., the estimated center position of the road). As described above, the road model outputs a plurality of position information representing the center of the road. Therefore, when these results are plotted, it does not form a single road, and as shown in the figure, missing parts or scattered parts (illustrated by broken lines) may occur. Since it cannot be used as a road map in this state, the estimation unit 113 executes a process to correct this.

[0056] First, the estimation unit 113 skeletonizes the image obtained by plotting the output of the road model. Skeletonization is a process of reducing the width of a line. As a result, for example, a line with a width of 2 pixels or more is corrected to a width of 1 pixel. Then, the estimation unit 113 extracts lines from the skeletonized image and vectorizes them.

[0057] (1) Gap interpolation process Next, the estimation unit 113 executes a gap interpolation process. The gap interpolation process refers to a process of connecting adjacent dead ends to each other (point to point), or a process of extending a dead end forward until it intersects an existing edge and then connecting it (point to edge). Note that it is preferable to set an upper limit to the length of gap interpolation.

[0058] Figure 4(C) is an example of a road graph after executing the gap interpolation process for the range surrounded by reference numeral 404. In this example, gap interpolation is performed at four locations indicated by reference numerals 405 to 408. Note that at the location indicated by reference numeral 405, both point to point and point to edge interpolations are performed.

[0059] (2) Edge deletion process Next, for each of the edges generated by the gap interpolation process, the estimation unit 113 determines the number of vehicles that passed through the road during a past predetermined period. The number of vehicles that passed through a certain edge can be determined based on the vehicle data stored in the storage unit 12. Here, an edge where the number of vehicles passing through during the past predetermined period is less than a predetermined threshold is determined not to be a road and is deleted. FIG. 4(D) is an example of a road graph after deleting the edges where the number of vehicles passing through is less than the predetermined threshold. In this example, the interpolated edges other than the portion indicated by reference numeral 409 are deleted.

[0060] Next, the estimation unit 113 deletes the remaining dead-end edges from the road graph. Here, an edge whose length is less than a predetermined threshold is a deletion target. FIG. 4(E) is an example of a road graph after deleting the dead-end edges.

[0061] By the processes described above, missing edges and jumping edges can be corrected, and a road graph representing only the portions where roads exist can be obtained.

[0062] (3) Intersection integration process Next, the estimation unit 113 executes an intersection integration process. In the result of plotting the output from the road model, at the locations where roads intersect, the edges do not always intersect at a single point. For example, when the road graph is as shown in FIG. 4(F), although there is one intersection, it may be determined to be two intersections. Therefore, when there are two or more intersections whose intervals are less than a predetermined threshold (for example, when two or more intersections occur within a range of 10 m radius), the estimation unit 113 executes a process of integrating these into one. FIG. 4(G) is an example of a road graph after performing the intersection integration process.

[0063] (4) Intersection determination process In the above-described process, it was assumed that intersections exist at the points where the edges intersect, but there may be overpasses or underpasses instead of intersections. Therefore, the estimation unit 113 executes a process of estimating the existence of overpasses or underpasses at the locations where the edges intersect.

[0064] For example, consider a case where there is a road graph as shown in FIG. 4(G). In this example, nodes are arranged assuming intersections, but there is also a possibility that two edges intersect three-dimensionally. Therefore, the estimation unit 113 takes statistics on the traveling directions of vehicles based on the vehicle data stored in the storage unit 12 for the target point. When the target point is an intersection, as shown in FIG. 4(H), vehicles traveling from a specific direction can diverge and travel in two or more directions at the said point. That is, when there are a plurality of traffic flows traveling toward the intersecting edges at the said point, it is estimated that there is an intersection at the said point. On the other hand, when there is no intersection at the said point, for example, in the case of an overpass or an underpass, there is no traffic flow traveling toward the intersecting edges. In this way, the estimation unit 113 determines whether each of the plurality of vehicles that have passed the target point in the past has changed its traveling direction at the target point, and based on the determination result, determines whether there is an intersection at the target point. When it is determined that there is an intersection at the target point, a node is arranged at the intersection of the road graphs. When it is determined that there is no intersection at the target point, no node is arranged at the intersection of the road graphs.

[0065] By performing the correction process described above, the estimation unit 113 can convert the set of points output by the road model into a road graph.

[0066] [Processing Flow] Next, the flow of the process executed by the server device 10 will be described. FIG. 5 is a flowchart of the process executed by the server device 10. Before the illustrated process starts, it is assumed that the server device 10 (acquisition unit 111) acquires vehicle data from a plurality of vehicles 1 under its management and stores it in the storage unit 12.

[0067] FIG. 5(A) is a flowchart of a process for the server device 10 to learn a road model. First, in step S11, the acquisition unit 111 acquires a master map from the map server 20. The master map is teacher data for learning the road model. The master map does not necessarily have to include the entire area for which the server device 10 generates a road graph. The acquired master map is stored in the storage unit 12.

[0068] Next, in step S12, the learning unit 112 acquires first vehicle data. The first vehicle data is input data used when learning the road model. The first vehicle data is preferably collected from the vehicle 1 that has traveled within the area of the master map.

[0069] Next, in step S13, the learning unit 112 executes learning of the road model. The learning of the road model may be performed for each unit area. For example, the master map is divided into a plurality of unit areas, and the position of the center line of the road included in the unit area is extracted as teacher data. Then, a set of position information included in the vehicle data of a plurality of vehicles 1 that have a record of traveling in the unit area is extracted as input data. Then, based on the extracted teacher data and input data, learning of the road model is executed. Note that it is preferable to align the relative positional relationships of the teacher data and the input data with each other.

[0070] FIG. 5(B) is a flowchart of a process for the server device 10 to generate a road graph using the learned road model. The illustrated process starts at an arbitrary timing after the road model is generated.

[0071] First, in step S21, the estimation unit 113 acquires second vehicle data. In the present embodiment, the estimation unit 113 acquires the second vehicle data collected within the area for generating the road graph from the storage unit 12. The area for generating the road graph may be an area not included in the master map.

[0072] Next, in step S22, the estimation unit 113 inputs the set of position information included in the second vehicle data acquired in step S21 into the road model, and acquires the set of position information output from the road model. The output set of position information is for estimating the position of the center line of the road.

[0073] Next, in step S23, the estimation unit 113 performs the above-described correction process based on the acquired set of position information to generate a road graph. FIG. 5(C) is a flowchart showing in detail the correction process executed in step S23. First, in step S231, the set of position information is converted into vector data. For example, the estimation unit 113 generates an image plotting the set of position information output by the road model, extracts a line from the image, and vectorizes it.

[0074] Next, in step S232, the above-described gap interpolation process and edge deletion process are executed. In this step, processes such as connecting the ends of edges that are close to each other and extending the ends of the edges until they intersect with other edges are executed. Also, in this step, the number of vehicle passages on the interpolated edge is determined, and an edge with the number of vehicle passages below a predetermined threshold is deleted. Such an edge may have been interpolated due to misjudgment. This is because the possibility is high. Also, the estimation unit 133 may delete the remaining dead-end edges after the process is completed.

[0075] Next, in step S233, intersection integration processing is performed. In this step, for example, when two or more points where edges intersect occur within a predetermined distance, they are integrated. Then, in step S234, an intersection determination process is performed. In this step, for each of the edges that intersect with each other, by obtaining the inflow traffic volume and the outflow traffic volume, it is determined whether it is possible to proceed in different directions. For example, when two edges intersect, it is determined whether the incoming traffic is proceeding on a specific edge or is divided and proceeding on two or more edges. If it is possible to proceed in different directions, it is determined that there is an intersection at the target point, and a node is arranged at the intersection point. If it is not possible to proceed in different directions, it is determined that there is a grade separation at the target point, and no node is arranged at the intersection point.

[0076] After step S23 is completed, the server device 10 may output the generated road graph to the outside of the device. Also, when the road graph is generated for each unit area, a plurality of road graphs generated for each unit area may be integrated.

[0077] Furthermore, the estimation unit 113 may perform a process of processing the road graph in order to generate road map data. Since the road graph generated in this embodiment represents the center line of the road, it does not have information such as the width of the road. Therefore, a process of adding these information may be executed. For example, for each edge, by determining the amount of traffic per unit time, the width of the road and the number of lanes can be estimated. For example, it may be determined that the wider the edge with more traffic volume is. In this case, for each edge, the traffic volume and the estimated road width may be associated.

[0078] As described above, the server device 10 according to the first embodiment learns the relative positional relationship between the GPS position information of the probe car and the road (for example, the center line of the road) by the machine learning model (road model). Since the position information by GPS includes errors, it is difficult to determine the accurate position of the road as it is. However, according to this embodiment, based on the set of position information of any vehicle, it becomes possible to accurately estimate the position of the road on which the vehicle has traveled.

[0079] (Second Embodiment) In the first embodiment, as input to the road model, the GPS position information acquired by vehicle 1 was used. However, the target road is not necessarily a single lane, and the width of the road varies. For example, even on the same road, vehicle 1 may be driving in the leftmost lane or the rightmost lane. If all of these are associated with the center line of the road and learned, the estimation accuracy may decrease. To address this problem, it is preferable to include in the input data to the road model not only the position information of vehicle 1 but also information indicating "where on the road the vehicle is driving".

[0080] In the second embodiment, as input data to the road model, in addition to the position information of vehicle 1 acquired by GPS, the following two types are added. (1) Information regarding the position of the road boundaries as seen from vehicle 1 (2) Information regarding the position of the lane markings as seen from vehicle 1 In the second embodiment, vehicle 1 has an in-vehicle camera mounted facing forward. Also, the in-vehicle device 30 has a function of detecting the positions of the road boundaries and the lane markings based on the image acquired by the in-vehicle camera. The road boundaries are lines that divide the road from the outside of the road, and the lane markings are lines that demarcate the lanes (for example, white lines or broken lines). Note that the road boundaries may be virtual lines representing the ends of the road area and do not necessarily have to be lines marked on the road surface.

[0081] FIG. 7 is a diagram showing an example of the configuration of vehicle 1 in the second embodiment. Vehicle 1 according to the second embodiment is further configured to include a camera 35. Also, the control unit 31 is further configured to include an image analysis unit 312 as a software module.

[0082] The camera 35 is an image sensor mounted facing the front of the vehicle 1. The camera 35 can capture an image of the front of the vehicle and transmit the image to the image analysis unit 312. The image analysis unit 312 analyzes the image captured by the camera 35 to detect the positions of the road boundary line and the lane boundary line in front of the vehicle 1. The image analysis unit 312, for example, detects a solid white line or a broken line from within the image and determines the relative position of the lane boundary line with respect to the host vehicle. Also, the image analysis unit 312 detects the edge of the road area and determines the relative position of the road boundary line with respect to the host vehicle.

[0083] FIG. 8 is a diagram for explaining the information included in the vehicle data generated in the second embodiment. In the second embodiment, the vehicle data includes the position information of the host vehicle, information regarding the relative position of the lane boundary line as seen from the host vehicle (relative position information), and the relative position information of the road boundary line. The relative position information of the lane boundary line and the road boundary line corresponds to the range captured by the camera 35. Note that the position information of the lane boundary line and the road boundary line may be represented by a set of points or by a vector line.

[0084] FIG. 5(D) is a flowchart of the process in which the vehicle 1 (message transmission unit 311) transmits vehicle data in the second embodiment. This process is periodically executed while the vehicle 1 is running.

[0085] First, in step S41, the position information of the host vehicle is acquired via the position information acquisition unit 34. Next, in step S42, the image analysis unit 312 is used to analyze the image captured by the camera 35 to acquire the relative position information of the lane boundary line as seen from the host vehicle. Also, in step S43, the image analysis unit 312 is used to analyze the image captured by the camera 35 to acquire the relative position information of the road boundary line as seen from the host vehicle. Then, in step S44, vehicle data including the position information of the host vehicle, and the relative position information of the road boundary line and the lane boundary line is generated and transmitted to the server device 10.

[0086] The server device 10 specifies the absolute positions of the road boundary line and the lane boundary line based on the relative position information of the road boundary line / lane boundary line included in the first vehicle data. Further, in addition to these position information being added to the input data, learning of the road model is executed. FIG. 9(A) is a diagram for explaining the learning phase in the second embodiment. As illustrated, in the second embodiment, as input data, in addition to the first vehicle data, information regarding the position of the road boundary line (for example, the position of the road boundary line represented by absolute coordinates), and information regarding the position of the lane boundary line (for example, the position of the lane boundary line represented by absolute coordinates) are used. Thereby, learning of the road model can be performed in consideration of information such as "where the vehicle 1 is located within the road". The same applies to the estimation phase. FIG. 9(B) is a diagram for explaining the estimation phase in the second embodiment. The server device 10 specifies the absolute positions of the road boundary line and the lane boundary line based on the relative position information of the road boundary line and the lane boundary line included in the second vehicle data. Further, in addition to these position information being added to the input data, estimation is executed. According to such a configuration, even when the road width is wide or when there are multiple lanes, the accuracy of the estimation can be maintained.

[0087] In this embodiment, the in-vehicle device 30 detects the road boundary line and the lane boundary line and transmits the results to the server device 10. However, as long as it can indicate "where the host vehicle is located within the road", the in-vehicle device 30 may transmit other information. For example, the in-vehicle device 30 may make a determination such as "which lane the host vehicle is traveling in" and include this information in the vehicle data and transmit it to the server device 10.

[0088] (Modification example) The above embodiment is merely an example, and the present disclosure can be appropriately modified and implemented without departing from the gist thereof. For example, the processes and means described in the present disclosure can be freely combined and implemented as long as no technical contradiction occurs.

[0089] Also, in the embodiment, it is assumed that the master map defines the center line of the road. However, the master map does not necessarily have to define the position of the center line of the road as long as it indicates the position of the road.

[0090] Also, the processing described as being performed by one device may be shared and executed by a plurality of devices. Alternatively, the processing described as being performed by different devices may be executed by one device. In a computer system, it is possible to flexibly change how each function is realized by a hardware configuration (server configuration).

[0091] The present disclosure can also be realized by supplying a computer program that implements the functions described in the above embodiments to a computer and causing one or more processors included in the computer to read and execute the program. Such a computer program may be provided to the computer by a non-transitory computer-readable storage medium connectable to the system bus of the computer, or may be provided to the computer via a network. The non-transitory computer-readable storage medium includes, for example, any type of disk such as a magnetic disk (e.g., a floppy (registered trademark) disk, a hard disk drive (HDD), etc.), an optical disk (e.g., a CD-ROM, a DVD disk, a Blu-ray disk, etc.), a read-only memory (ROM), a random access memory (RAM), an EPROM, an EEPROM, a magnetic card, a flash memory, an optical card, and any type of medium suitable for storing electronic instructions.

Description of Reference Numerals

[0092] 1 ··· Vehicle 10 ··· Server device 11 ··· Control unit 12 ··· Storage unit 13 ··· Communication unit 20 ··· Map server 30 ··· In-vehicle device

Claims

1. Obtaining first map data including road position information; Obtaining probe data including a set of position information of the first moving body, which is measured by the first moving body moving on the road; Training a machine learning model using the first map data as teacher data and the probe data as input data; Converting second probe data including a set of position information of a second moving body into a road graph including road position information using the trained machine learning model; An information processing apparatus having a control unit that executes the above.

2. The control unit causes the machine learning model to learn the relative positional relationship between the set of position information of the first moving body and the actual road. The information processing apparatus according to Claim 1.

3. The first map data includes position information of the center line of the road. The control unit trains the machine learning model using the position information of the center line of the road included in the first map data as teacher data. The information processing apparatus according to Claim 1.

4. The control unit inputs the set of position information included in the second probe data into the trained machine learning model and obtains a set of position information of the center line of the road as an estimation result. The information processing apparatus according to Claim 3.

5. The probe data further includes data regarding the positions of road boundaries and / or lane boundaries. The control unit further includes the data in the input data to train the machine learning model. The information processing apparatus according to Claim 1.

Images