Information processing device

Abstract

To provide an information processing device capable of estimating a traffic volume of all vehicles traveling on an arbitrary road.SOLUTION: An information processing device has a learned model which is machine-learned to output a whole traffic volume of all vehicles 20 including connected vehicles 22 at an observation point with respect to a traffic volume and movement tendency of the connected vehicles 22 having traveled at the observation point where the traffic volume of traveling vehicles 20 is actually measured and also with respect to road information at the observation point. The device performs: acquiring vehicle information of the connected vehicles 22 having traveled at a non-observation point where the traffic volume of traveling vehicles is not measured and road information at the non-observation point; inputting the acquired vehicle information of the connected vehicles 22 having traveled at the non-observation point and the road information at the non-observation point to the learned model; and then estimating the whole traffic volume of all vehicles 20 at the non-observation point.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to an information processing apparatus.

Background Art

[0002] Patent Document 1 discloses a traffic volume prediction apparatus using a machine learning algorithm that outputs a predicted traffic volume based on the traffic volume by a traffic counter and the position information of vehicles and probe data (specifically, vehicle speed, traveling direction, etc.).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] The apparatus of Patent Document 1 learns the relationship between speed and traffic volume at a specific road, that is, the installation point of the traffic counter, and it is difficult to predict the traffic volume on an arbitrary road.

[0005] An object of the present invention is to provide an information processing apparatus capable of estimating the traffic volume of all vehicles traveling on an arbitrary road.

Means for Solving the Problems

[0006] The information processing device according to claim 1 includes an acquisition unit that acquires vehicle information of connected vehicles capable of wireless communication that have traveled to non-observation points where the traffic volume of moving vehicles is not measured, and road information of the non-observation points, and an estimation unit that estimates the total traffic volume of all vehicles at the non-observation points by inputting the acquired vehicle information of the connected vehicles that traveled to the non-observation points and the road information of the non-observation points to a trained model that has been trained by machine learning to output the total traffic volume of all vehicles, including the connected vehicles, at the observation points, based on the traffic volume and movement trends of the connected vehicles that have traveled to observation points where the traffic volume of moving vehicles is measured, and the road information of the observation points.

[0007] The information processing device described in claim 1 has a pre-trained model that has been trained by machine learning to output the total traffic volume of all vehicles, including connected vehicles, at an observation point, based on the traffic volume and movement trends of connected vehicles that have traveled to an observation point where the traffic volume of moving vehicles is actually measured, as well as road information of the observation point. The information processing device takes vehicle information of connected vehicles that have traveled to a non-observation point where the traffic volume of moving vehicles is not actually measured, and road information of the non-observation point as input to the pre-trained model, and estimates the total traffic volume of all vehicles at the non-observation point. Therefore, the information processing device can estimate the traffic volume of all vehicles traveling on any road.

[0008] The information processing device according to claim 2 is the information processing device according to claim 1, wherein the movement trend of the connected vehicle is information including at least one of the movement speed of the connected vehicle, the rate of branching from the road on which it is traveling, and lane change information on the road on which it is traveling.

[0009] According to the information processing device described in claim 2, machine learning is performed based on the specific movement trends of connected vehicles, thereby improving the accuracy of traffic volume estimation on any road.

[0010] The information processing device according to claim 3 is the information processing device according to claim 1, wherein the road information includes at least one of the number of lanes on the road on which the vehicle travels, the number of other roads from which the road connects, the number of other roads to which the road connects, the speed limit of the road, and geographical information of the road.

[0011] According to the information processing device described in claim 3, machine learning is performed based on specific road information, and estimation is performed based on specific road information, thereby improving the accuracy of traffic volume estimation on any road.

[0012] The information processing device according to claim 4 is an information processing device according to any one of claims 1 to 3, wherein vehicle information of the connected vehicle can be acquired at unit time intervals, and the unit time interval is shorter than the interval at which the traffic volume of the vehicle at the observation point is measured.

[0013] In the information processing device described in claim 4, even if the interval between measuring the actual traffic volume of vehicles at the observation point is long, vehicle information of connected vehicles is acquired at unit time intervals shorter than the said interval, and traffic volume is estimated. Therefore, with this information processing device, it is possible to grasp the traffic volume on any road in real time compared to when traffic volume is measured using traffic counters or manually.

[0014] The information processing device according to claim 5 includes: an acquisition unit that acquires vehicle information of connected vehicles capable of wireless communication among vehicles that have traveled to an observation point where the traffic volume of moving vehicles is actually measured, road information of the observation point, and the traffic volume of all vehicles, including the connected vehicle, at the observation point; an analysis unit that analyzes the traffic volume and movement trends of the connected vehicle based on the acquired vehicle information, and the total traffic volume of all vehicles at the observation point based on the acquired traffic volume; and a learning unit that generates a trained model that outputs the total traffic volume of all vehicles at the non-observation point based on the input of vehicle information of the connected vehicle that has traveled to a non-observation point where the traffic volume of moving vehicles is not actually measured, by performing machine learning to output the analyzed total traffic volume with respect to the analyzed traffic volume and movement trends, and the acquired road information of the observation point.

[0015] The information processing device described in claim 5 performs machine learning to output the total traffic volume of all vehicles, including connected vehicles, at an observation point, based on the traffic volume and movement trends of connected vehicles that traveled to an observation point where the traffic volume of moving vehicles is actually measured, as well as road information of the observation point, and generates a trained model. The information processing device can estimate the total traffic volume of all vehicles at an observation point by inputting vehicle information of connected vehicles that traveled to an observation point where the traffic volume of moving vehicles is not actually measured, and road information of the observation point, into the generated trained model. Therefore, the information processing device can estimate the traffic volume of all vehicles traveling on any road. [Effects of the Invention]

[0016] According to the present invention, it is possible to estimate the total traffic volume of all vehicles traveling on any given road. [Brief explanation of the drawing]

[0017] [Figure 1] This figure shows the schematic configuration of the estimation system in the embodiment. [Figure 2] This is a block diagram showing the hardware configuration of the processing server. [Figure 3] It is a block diagram showing the configuration of the storage of the processing server. [Figure 4] It is a block diagram showing the functional configuration of the processing server. [Figure 5] It is a flowchart showing the flow of the learning process. [Figure 6] It is a flowchart showing the flow of the estimation process.

Embodiments for Carrying Out the Invention

[0018] Hereinafter, the estimation system 10 according to the present embodiment will be described. FIG. 1 is a diagram showing the schematic configuration of the estimation system 10.

[0019] (Configuration) As shown in FIG. 1, the estimation system 10 includes a plurality of vehicles 20, a processing server 30, and an information providing server 32. The vehicles 20 include connected vehicles 22 capable of communicating with external devices such as the processing server 30 and non-connected vehicles 24 incapable of communicating with external devices. The number of connected vehicles 22 and the number of non-connected vehicles 24 are not limited to the numbers shown in FIG. 1. Each connected vehicle 22, processing server 30, and information providing server 32 are interconnected via a network N.

[0020] The connected vehicle 22 is a so-called connected car equipped with a DCM (Data Communication Module) and capable of transmitting the vehicle information of its own vehicle to the processing server 30. The vehicle information includes, for example, identification information for identifying the connected vehicle 22, location information, and time information. The location information is information indicating the location of the connected vehicle 22. The time information is information indicating the date and time when the connected vehicle 22 acquired the location information. The connected vehicle 22 acquires location information at regular intervals, for example, by using a location information acquisition system such as GNSS (Global Navigation Satellite System). In addition, the connected vehicle 22 acquires, as time information, the date and time measured by an in-vehicle clock unit (not shown) when the location information is acquired. When the connected vehicle 22 acquires the location information and the time information, it stores them in a predetermined storage area and transmits them to the processing server 30 as vehicle information at a predetermined timing.

[0021] The processing server 30, which serves as an information processing device, has a function of generating an estimation model 110 described later for estimating traffic volume by machine learning, and a function of estimating the traffic volume of the vehicle 20 by using the estimation model 110. The details of the processing server 30 will be described later.

[0022] The information providing server 32 has at least a function of providing the processing server 30 with the traffic volume of at least past observation points. The traffic volume of past observation points provided by the information providing server 32 is measured manually or by a traffic counter. The traffic volume measured manually and the traffic volume measured by the traffic counter are data with a period of about 1 to several months from the actual measurement until it is generally made public. Note that the information providing server 32 may provide the processing server 30 not only with the traffic volume of past observation points but also with the vehicle information of the connected vehicle 22 and the road information of the road on which the vehicle 20 travels.

[0023] Next, the hardware configuration of the processing server 30 will be described. FIG. 2 is a block diagram showing the hardware configuration of the processing server 30.

[0024] As shown in Figure 2, the processing server 30 includes a CPU (Central Processing Unit) 30A, ROM (Read Only Memory) 30B, RAM (Random Access Memory) 30C, storage 30D, and a communication interface (I / F) 30E. Each component is connected to the others via a bus 30G so that they can communicate with each other.

[0025] The CPU 30A is the central processing unit, which executes various programs and controls various components. Specifically, the CPU 30A reads programs from ROM 30B or storage 30D and executes them using RAM 30C as the working area. The CPU 30A controls each component and performs various calculations according to the programs recorded in ROM 30B or storage 30D.

[0026] ROM30B stores various programs and data. RAM30C temporarily stores programs or data as a working area.

[0027] Storage 30D is composed of a storage device such as an HDD (Hard Disk Drive), SSD (Solid State Drive), or flash memory, and stores various programs and data. As shown in Figure 3, Storage 30D stores the processing program 100, the estimation model 110, the vehicle information database 120, the road information database 130, and the traffic volume database 140. The processing program 100, the estimation model 110, the vehicle information database 120, the road information database 130, and the traffic volume database 140 may also be stored in ROM 30B.

[0028] The processing program 100 is a program that performs the learning process and estimation process described later. Estimated model 110 is a pre-trained model generated by machine learning.

[0029] The vehicle information database 120 stores vehicle information collected from the connected vehicle 22, and movement trends analyzed based on the vehicle information. Here, the vehicle information includes at least the location information and time information of the connected vehicle 22 as described above. In addition to location information and time information, the vehicle information may also include information that can be collected by the connected vehicle 22, such as speed, acceleration, steering angle, etc. The movement information includes the speed of the connected vehicle 22, the probability of branching from the road it is traveling on (for example, the probability of turning right, turning left, or going straight), and information related to lane changes on the road it is traveling on. The movement trends only need to include at least one of the speed of travel, branching probability, and lane change information.

[0030] The road information database 130 stores road information for the roads on which vehicle 20 travels. This road information includes the number of lanes on the road, the number of other roads to which the road connects, the number of other roads to which the road connects, the speed limit, and geographical information of the road. Note that the road information only needs to include at least one of the following: the number of lanes, the number of other roads to which the road connects, the number of other roads to which the road connects, the speed limit, and geographical information of the road. Furthermore, the information included in the road information may differ for each road.

[0031] The traffic volume database 140 stores information on the traffic volume measured at observation points where the traffic volume of vehicles 20 is actually measured. Examples of methods for measuring the traffic volume of vehicles 20 include using traffic counters installed on the road and measuring manually. According to these methods, the traffic volume at observation points is updated at regular intervals (for example, every one to several months). The traffic volume stored in the traffic volume database 140 is provided by the information provision server 32. The traffic volume provided is also provided as information by time and by vehicle type. Note that locations where the traffic volume of vehicles 20 is not measured are referred to as non-observation points, and the two are distinguished from observation points.

[0032] As shown in Figure 2, the communication I / F30E is an interface for connecting to network N. Communication via the communication I / F30E can use, for example, wired communication standards such as Ethernet® or FDDI, or wireless communication standards such as 4G, 5G, Bluetooth®, or Wi-Fi®.

[0033] Next, the functional configuration of the processing server 30 will be described. As shown in Figure 4, the CPU 30A of the processing server 30 has the following functional configuration: acquisition unit 200, analysis unit 210, learning unit 220, estimation unit 230, and provision unit 240. Each functional configuration is realized by the CPU 30A reading and executing the processing program 100 stored in the storage 30D.

[0034] The acquisition unit 200 has the function of acquiring various information from the connected vehicle 22, the information provision server 32, and the storage 30D. In the learning process described later, the acquisition unit 200 of this embodiment acquires vehicle information from at least the connected vehicle 22 that has traveled to the observation point and acquires road information at the observation point. The acquisition unit 200 also acquires the traffic volume of the vehicle 20 measured at the observation point from the information provision server 32.

[0035] Furthermore, in the estimation process described later, the acquisition unit 200 acquires vehicle information of at least connected vehicles 22 that have traveled to non-observation points and acquires road information of those non-observation points. In the estimation process, the acquisition unit 200 acquires vehicle information from connected vehicles 22 at predetermined intervals (for example, several minutes to several hours).

[0036] The analysis unit 210 has the function of analyzing the traffic volume of connected vehicles 22 and the movement trends of said connected vehicles 22 based on the vehicle information acquired by the acquisition unit 200. In addition, the analysis unit 210 analyzes the traffic volume of all vehicles 20, including connected vehicles 22, at the observation point (hereinafter sometimes referred to as "total traffic volume") based on the measured traffic volume acquired by the acquisition unit 200. The total traffic volume is the traffic volume that includes the traffic volume of each vehicle type at the observation point.

[0037] The learning unit 220 has the function of generating an estimation model 110 by performing machine learning through a learning process described later. In this embodiment, the learning unit 220 performs machine learning based on training data in which the total traffic volume analyzed by the analysis unit 210 is associated with the traffic volume and movement trends of connected vehicles 22 analyzed by the analysis unit 210, and the road information of observation points acquired by the acquisition unit 200.

[0038] The estimation unit 230 has the function of estimating traffic volume at observation points and non-observation points using the estimation model 110. In this embodiment, the estimation unit 230 estimates the total traffic volume of all vehicles 20 at non-observation points by inputting vehicle information of connected vehicles 22 that traveled through non-observation points acquired by the acquisition unit 200, and road information of those non-observation points, to the estimation model 110.

[0039] The data provision unit 240 has the function of providing the total traffic volume estimated by the estimation unit 230 to an external device. Examples of recipients of the total traffic volume provided by the data provision unit 240 include the car navigation system of each vehicle 20 and mobile devices such as smartphones held by the occupants of the vehicle 20.

[0040] (Control flow) The processing flow performed in the estimation system 10 of this embodiment will be explained using the flowcharts in Figures 5 and 6. Each process in the processing server 30 is performed by the CPU 30A functioning as an acquisition unit 200, an analysis unit 210, a learning unit 220, an estimation unit 230, and a provision unit 240.

[0041] First, let's explain the learning process that generates the estimated model 110. In step S100 of Figure 5, the CPU 30A acquires vehicle information for the connected vehicle 22 on an hourly basis.

[0042] In step S102, the CPU 30A analyzes the traffic volume and movement trends of connected vehicles 22 per unit time at past observation points based on the acquired vehicle information.

[0043] In step S104, CPU30A acquires the traffic volume at the observation point.

[0044] In step S106, CPU30A analyzes the total traffic volume per unit time at past observation points.

[0045] In step S108, the CPU 30A acquires road information for the road including the observation point.

[0046] In step S110, the CPU 30A generates an estimation model 110 using machine learning. Specifically, the CPU 30A performs machine learning based on training data that associates the total traffic volume analyzed in step S106 with the traffic volume and movement trends of connected vehicles 22 analyzed in step S102, and the road information of observation points acquired in step S108. This generates the estimation model 110, and the learning process is then completed.

[0047] Next, we will describe the estimation process for estimating the total traffic volume using estimation model 110. In step S200 of Figure 6, the CPU 30A acquires vehicle information for all connected vehicles 22 on the roads to be estimated.

[0048] In step S202, the CPU 30A acquires road information for all roads to be estimated.

[0049] In step S204, the CPU 30A estimates the total traffic volume using the estimation model 110. Specifically, the CPU 30A inputs the vehicle information of the connected vehicles 22 that traveled on the road to be estimated, which was obtained in step S200, and the road information of the road to be estimated, which was obtained in step S202, to the estimation model 110. This estimates the total traffic volume of all vehicles 20 on the road to be estimated.

[0050] In step S206, the CPU 30A provides the total traffic volume estimated in step S204 to an external device. Specifically, the CPU 30A transmits the total traffic volume to the car navigation system of the vehicle 20 or to a smartphone held by an occupant of the vehicle 20. The estimation then ends.

[0051] (summary) The processing server 30 of this embodiment has an estimation model 110 that has been trained by machine learning to output the total traffic volume of all vehicles 20, including connected vehicles 22, at an observation point, based on the traffic volume and movement trends of connected vehicles 22 that have traveled at the observation point, as well as road information of the observation point. The processing server 30 inputs vehicle information of connected vehicles 22 that have traveled on the road to be estimated and road information of the road to be estimated to the estimation model 110, and estimates the total traffic volume of all vehicles 20 on the road to be estimated. Therefore, according to the processing server 30 of this embodiment, it is possible to estimate the traffic volume of all vehicles traveling on any road.

[0052] In particular, according to this embodiment, when estimating traffic volume at non-observed locations, accurate estimation is possible when there are observation locations with road information and movement trends similar to those of the non-observed location being estimated. Furthermore, according to this embodiment, by using more traffic volume data from observation locations throughout the country, vehicle information from connected vehicles 22, and road information data, highly accurate traffic volume estimation becomes possible. And, since it is possible to estimate traffic volume at non-observed locations based on data from observation locations, the cost of manually measuring traffic volume is reduced.

[0053] Here, the movement trend of the connected vehicle 22 is information including at least one of the movement speed of the connected vehicle 22, the rate of branching off from the road it is traveling on, and lane change information on the road it is traveling on. Therefore, according to the processing server 30 of this embodiment, machine learning is performed based on the specific movement trend of the connected vehicle 22, which improves the accuracy of traffic volume estimation on any road.

[0054] Furthermore, the road information includes at least one of the following: the number of lanes on the road on which the vehicle 20 travels, the number of other roads to which the road connects, the number of other roads to which the road connects, the speed limit of the road, and geographical information of the road. Therefore, according to the processing server 30 of this embodiment, machine learning is performed based on specific road information, and estimation is performed based on specific road information, thereby improving the accuracy of traffic volume estimation on any road.

[0055] The processing server 30 of this embodiment can acquire vehicle information of connected vehicles 22 at unit time intervals (e.g., several minutes to several hours). This unit time interval is shorter than the interval at which the traffic volume of vehicles 20 at the observation point is measured (e.g., one to several months). Therefore, according to this embodiment, even if the interval at which the traffic volume of vehicles 20 at the observation point is measured is a long period of time of one to several months, vehicle information of connected vehicles 22 is acquired at unit time intervals shorter than that interval, and traffic volume is estimated. Therefore, according to the processing server 30 of this embodiment, it is possible to grasp the traffic volume on any road in real time compared to when traffic volume is measured by traffic counters or manually.

[0056] [remarks] In the above embodiment, the various processes that the CPU 30A reads and executes software (programs) may be executed by various processors other than the CPU. Examples of such processors include PLDs (Programmable Logic Devices) such as FPGAs (Field-Programmable Gate Arrays) whose circuit configuration can be changed after manufacturing, and dedicated electrical circuits that are processors with circuit configurations specifically designed to execute specific processes, such as ASICs (Application Specific Integrated Circuits). Furthermore, each of the above processes may be executed by one of these various processors, or by a combination of two or more processors of the same or different types (for example, multiple FPGAs, and a combination of a CPU and an FPGA). More specifically, the hardware structure of these various processors is an electrical circuit that combines circuit elements such as semiconductor elements.

[0057] Furthermore, in the above embodiment, each program was described as being pre-stored (installed) on a computer-readable non-temporary recording medium. For example, processing program 100 is pre-stored on storage 30D. However, the program is not limited to this, and each program may be provided in a form recorded on a non-temporary recording medium such as a CD-ROM (Compact Disc Read Only Memory), DVD-ROM (Digital Versatile Disc Read Only Memory), or USB (Universal Serial Bus) memory. Alternatively, the program may be downloaded from an external device via a network.

[0058] The processing flow described in the above embodiment is just one example, and unnecessary steps may be deleted, new steps added, or the processing order rearranged, as long as it does not deviate from the main point. [Explanation of Symbols]

[0059] 20 vehicles 22 Connected Vehicles 30. Processing Server (Information Processing Device) 110 Estimation Model (Pre-trained Model) 200 Acquisition Department 210 Analysis Department 220 Learning Department 230 Estimation Department

Claims

[Claim 1] An acquisition unit that acquires vehicle information of connected vehicles capable of wireless communication among vehicles that have traveled to an observation point where the traffic volume of moving vehicles is measured, road information of the observation point, and the traffic volume of all vehicles, including the connected vehicle, at the observation point. An analysis unit analyzes the traffic volume and movement trends of the connected vehicles based on the acquired vehicle information, and the total traffic volume of all vehicles at the observation point based on the acquired traffic volume. A learning unit generates a trained model that outputs the total traffic volume of all vehicles at non-observation points based on the input of vehicle information of connected vehicles that traveled at non-observation points where the traffic volume of vehicles traveling is not measured, by performing machine learning to output the analyzed total traffic volume with respect to the analyzed traffic volume and movement trends, and the road information of the acquired observation points. An information processing device equipped with the following features.

Images