Precipitation level estimation system and program

Abstract

To provide a precipitation level estimation system and program that can acquire data related to precipitation acquired from vehicles and accurately calculate a precipitation level using an estimation method based on machine learning.SOLUTION: A precipitation level estimation system 1 includes a calculation section 12 that acquires detection data related to precipitation, which is detected by a plurality of vehicles Mn present in an observation target area RX, and calculates a precipitation amount in a predetermined area Rm included in the observation target area on the basis of the detection data. The calculation section calculates a first feature amount in a first predetermined period, which is past with respect to present, on the basis of first detection data acquired from a first area including the predetermined area, calculates a second feature amount in a second predetermined period, which is past with respect to the first predetermined period, in a second area on the basis of second detection data acquired in the second area adjacent to the first area, calculates the precipitation amount using a precipitation amount estimation model using the first feature amount and the second feature amount as variables, and calculates a precipitation level according to the precipitation amount.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a precipitation level estimation system and program capable of estimating precipitation.

Background Art

[0002] In recent years, the disaster prevention awareness regarding the occurrence of local heavy rain has been increasing. Patent Document 1 describes a technique for calculating an index indicating the intensity of precipitation in an area where a vehicle is located based on data related to the operation mode of a wiper device of the vehicle.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] The operation mode of the wiper is provided with, for example, three operation modes such as weak, medium, and strong. According to the technique described in Patent Document 1, the intensity of precipitation is estimated based on data related to the operation mode of the wiper. The operation mode of the wiper is fixed at the strong stage particularly when heavy precipitation is occurring, and the relevance between the operation mode of the wiper and the precipitation amount decreases. Therefore, according to the technique described in Patent Document 1, there may be a difficulty in accurately estimating the precipitation level in a certain area.

[0005] An object of the present invention is to provide a precipitation level estimation system and program that can accurately calculate the precipitation level by using data related to precipitation acquired from a vehicle and using an estimation method based on machine learning.

Means for Solving the Problems

[0006] One aspect of the present invention is a precipitation level estimation system comprising: a calculation unit that acquires detection data related to precipitation detected by a plurality of vehicles present in an observation target area, and calculates the amount of precipitation in a predetermined area included in the observation target area based on the detection data, the calculation unit calculates a first feature quantity for a first predetermined period in the past compared to the present based on first detection data acquired from a first area including the predetermined area, a second feature quantity for a second predetermined period in the second area in the past compared to the first predetermined period based on second detection data acquired in a second area adjacent to the first area, calculates the amount of precipitation using a precipitation estimation model with the first feature quantity and the second feature quantity as variables, and calculates a precipitation level corresponding to the amount of precipitation. [Effects of the Invention]

[0007] According to the present invention, precipitation levels can be accurately calculated using a machine learning-based estimation method that utilizes precipitation-related data acquired from vehicles. [Brief explanation of the drawing]

[0008] [Figure 1] This is a block diagram showing the configuration of a precipitation level estimation system according to an embodiment. [Figure 2] This diagram shows the observation method in the area under observation. [Figure 3] This figure shows a designated area within the observation target region. [Figure 4] This diagram shows the contents of multiple features. [Figure 5] This figure shows mapping data with precipitation levels displayed. [Figure 6] This figure shows the changes in precipitation levels over time in the mapping data. [Figure 7] This flowchart shows the processing flow of the precipitation level estimation method. [Modes for carrying out the invention]

[0009] As shown in Figure 1, the precipitation level estimation system 1 consists of a server device 10 connected to a network NW and multiple vehicles Mn (n: any natural number). The multiple vehicles Mn are located in the observation area RX, which is the area to be observed. The observation area RX is set, for example, as a rectangular area of ​​a predetermined size. The server device 10 communicates with the multiple vehicles Mn and acquires detection data related to precipitation. Based on the detection data, the server device 10 calculates the precipitation level in the observation area RX.

[0010] The precipitation level is set to three levels according to the amount of precipitation, for example, as described later. Based on the calculation results, the server device 10 provides information including the precipitation level to the vehicle Mn within the observation target area RX. The server device 10 may also provide information to terminal devices 20 connected to the network NW in addition to the vehicle Mn.

[0011] The terminal device 20 is comprised of, for example, an information processing terminal device such as a personal computer or a smartphone. The terminal device 20 includes, for example, a display unit 26 for displaying information. The display unit 26 is comprised of, for example, a display device such as a liquid crystal display. The terminal device 20 includes a communication unit 28 capable of communicating with a network NW. The communication unit 28 is comprised of a communication interface capable of connecting to the network NW. The terminal device 20 includes a control unit 22 that performs communication and information display processing.

[0012] The terminal device 20 includes a storage unit 24 in which data and programs necessary for control are stored. The control unit 22 is composed of at least one hardware processor such as a CPU (Central Processing Unit). The storage unit 24 is composed of non-temporary storage media such as a hard disk drive (HDD) or solid state disk (SSD).

[0013] The server device 10 consists of an arithmetic unit 12 that performs various calculations, a storage unit 14 that stores data and programs necessary for calculations, and a communication unit 16 that communicates with the network NW. The arithmetic unit 12 is composed of at least one hardware processor such as a CPU. The storage unit 14 is composed of non-temporary storage media such as a hard disk drive or solid-state disk. The communication unit 16 is composed of a communication interface that can connect to the network NW.

[0014] The calculation unit 12 acquires detection data related to precipitation detected by multiple vehicles Mn located in the observation area RX. Based on the detection data, the calculation unit 12 calculates the amount of precipitation in the observation area RX. The details of the calculation process of the calculation unit 12 will be described later. Based on the calculation results, the calculation unit 12 calculates the distribution of precipitation levels in the observation area RX. The calculation unit 12 provides information on the amount of precipitation to the vehicles Mn used by the user via the network NW.

[0015] Vehicle Mn is configured to communicate with server device 10 via network NW. Vehicle Mn is equipped with a detection unit MS that detects various data. The detection unit MS is composed of multiple vehicle equipment such as devices and sensors installed on vehicle Mn. The detection unit MS is composed of vehicle equipment capable of acquiring control signals necessary for control, current values ​​necessary for operation, and data necessary for controlling vehicle Mn. The detection unit MS is equipped with a wiper device MN that wipes the windows of vehicle Mn.

[0016] The wiper device MN is provided on the front window and rear window of the vehicle Mn. The wiper device MN operates based on, for example, three operating modes such as weak, medium, and strong. The operating modes increase in operating speed in the order of weak, medium, and strong. The operating speed is set, for example, by the number of wiping times per unit time (times / minute). The operating mode may be provided with a speed adjustment mode in which the number of wiping times per unit time can be arbitrarily adjusted. The wiper device MN operates based on the operation of the occupant of the vehicle Mn. The wiper device MN operates based on a control signal corresponding to the operation of the occupant. The control signal is set corresponding to the stepped operating mode and speed adjustment mode.

[0017] The wiper device MN may be provided with a rain sensor ME for detecting rainfall. The rain sensor ME is provided inside the front window and detects a change in the light transmittance of the front window. The wiper device MN automatically operates, for example, when rainfall is detected by the rain sensor ME. The wiper device MN operates with the operating mode automatically set according to the change in the signal detected by the rain sensor ME.

[0018] The detection unit MS is provided with a camera MC capable of imaging an image. The camera MC images the external environment of the vehicle Mn. The imaging data imaged by the camera MC is used, for example, for the control of the vehicle Mn. The detection unit MS is provided with a lidar device MD that emits laser waves. The lidar device MD receives the reflected wave of the emitted laser wave and detects an object existing around the vehicle Mn. The detection unit MS is provided with a position sensor MG for detecting the current position of the vehicle Mn. The position sensor MG is composed of, for example, a GPS (Global Positioning System) sensor.

[0019] The vehicle Mn is provided with a control unit MP that executes various controls. The control unit MP executes controls related to the running of the vehicle Mn. The control unit MP transmits data related to precipitation among the data detected by the detection unit MS to the server device 10. The control unit MP is composed of a hardware processor such as at least one CPU.

[0020] The vehicle Mn is provided with a notification unit ML that notifies various information. The notification unit ML is, for example, composed of a display device capable of displaying an image including the notification content. The notification unit ML may be composed of a speaker that outputs the notification content by voice. The vehicle Mn includes a communication unit MU that can be connected to the network NW. The communication unit MU is, for example, composed of a communication interface capable of wireless communication.

[0021] During running, the vehicle Mn acquires various data by the detection unit MS. The control unit MP controls the running of the vehicle Mn based on the acquired data. The control unit MP transmits data related to precipitation among the data acquired by the detection unit MS to the server device 10. The control unit MP transmits, for example, data related to the control of the wiper device MN to the server device 10. In the server device 10, the data is stored in the storage unit 14. The calculation unit 12 calculates the precipitation amount in the observation target area RX based on the data stored in the storage unit 14.

[0022] As shown in FIG. 2, the map data of the observation target area RX includes an area Rm (m: natural number). The area Rm is set, for example, as a square rectangular area of 1 km on each side. The size, position, and shape of the area Rm are arbitrarily set. The area Rm is set with an ID number corresponding to the position. The area Rm is managed by the server device 10. The server device 10 communicates with a plurality of vehicles Mn existing in the observation target area RX. In the server device 10, the calculation unit 12 acquires the position data of each vehicle Mn and the ID data for identifying the vehicle Mn.

[0023] As shown in Figure 3, the calculation unit 12 compares the ID data RD of area Rm with the data obtained from the vehicle Mn to recognize the vehicle Mn present in area Rm. The calculation unit 12 obtains precipitation-related detection data from multiple vehicle Mn present in area Rm. Precipitation-related detection data includes, for example, data on the operating status of the wiper device MN, vehicle ID data, vehicle position data, and time data. The calculation unit 12 obtains detection data from the vehicle Mn present in area Rm at predetermined intervals (for example, 1-minute intervals). The calculation unit 12 stores the obtained detection data in the storage unit 14.

[0024] The calculation unit 12 calculates the amount of precipitation in region Rm. Region Rm includes the user locations of users who receive precipitation information from the server device 10. User location information is provided to the server device 10 via the vehicle Mn the user is currently riding in or via a terminal device 20 the user is carrying. Based on the location information provided by the user, the calculation unit 12 calculates the amount of precipitation in region Rm, which includes the user's location.

[0025] As shown in Figure 4, the calculation unit 12 generates multiple feature quantities B to Y used for estimating precipitation at predetermined intervals (e.g., 10-minute intervals) based on the detected data. The multiple feature quantities B to Y are classified into 24 types of elements, such as the speed component and time component of the wiper device MN, the number component of the vehicle Mn, and the region component. The multiple feature quantities B to Y are set so that features appear in the calculation of precipitation at the user's location based on data indicating the operating state of the wiper device MN.

[0026] The calculation unit 12 generates a first feature quantity Ta (feature quantities B to M) based on first detection data obtained from a first region Ra included in the observation target region RX. The first region Ra is defined by multiple sample data collection regions, for example, the user's position and its surrounding locations. The multiple sample data collection regions are defined by, for example, a 1km square region Rm including the user's position and multiple 1km square surrounding regions including surrounding locations located at a predetermined distance from the user's position.

[0027] The surrounding locations include, for example, a location 5 km away from the user's position and a location 10 km away from the user's position. The calculation unit 12 collects first detection data for a first region Ra that includes the user's position and surrounding locations in any direction from the user's position. The direction in which data is collected is arbitrarily set based on the sample data collection method. The calculation unit 12 collects data for the first region Ra in the four cardinal directions of east, west, south, and north from the user's position. The calculation unit 12 may also collect data for a first region that includes any directional element, such as eight cardinal directions, by adding northeast, northwest, southwest, and southeast directions to the east, west, south, and north directions.

[0028] The calculation unit 12 acquires first detection data of the first region Ra in a first predetermined period t1 that is in the past compared to the present. The first predetermined period t1 is, for example, one hour in the past from the present time. The calculation unit 12 generates feature quantities B to M in the four cardinal directions, for example, based on the first detection data acquired in the first region Ra. The calculation unit 12 generates feature quantities B to M based on wiper operation data that indicates the operating status of a wiper device installed on a vehicle, for example.

[0029] The calculation unit 12 calculates, for example, feature quantities related to the wiper operating time ratio, which indicates the operating status of the wiper device MN. The wiper operating time ratio is the proportion of time the wiper device was operating within a predetermined period. The wiper operating time ratio includes feature quantities B, C, and D, which are calculated from the maximum operating time of the wiper device within a predetermined period (e.g., 10 minutes), and feature quantities E, F, and G, which are calculated from the sum of the operating time of the wiper device within a predetermined period (e.g., 60 minutes).

[0030] Features B, C, and D, which indicate precipitation levels, are sampled a total of six times per hour at 10-minute aggregation intervals. By sampling features B, C, and D in a short period of time and using their maximum values ​​over that hour, along with features E, F, and G, which represent the sum of the time the wiper device was activated over that hour, it is possible to accurately capture precipitation peak values ​​indicating heavy rainfall.

[0031] Feature quantities B and E are calculated within a 1km x 1km rectangular area including the user's location. Feature quantities C and F are calculated within a 1km x 1km rectangular area 5Rm (see Figure 2) including a point 5km away from the user's location in a predetermined direction. Feature quantities D and G are calculated within a 1km x 1km rectangular area 10Rm (see Figure 2) including a point 10km away from the user's location in a predetermined direction.

[0032] The above features indicating precipitation levels are calculated not only for the 1km square area of ​​area Rm, which includes the user's location, in the first area Ra, but also for areas 5Rm and 10Rm, which are separated from area Rm in four directions. The wider the sampling area that forms the basis for calculating the features according to the direction, the more vehicles Mn from which data can be collected increase. Furthermore, sampling according to direction makes it possible to capture precipitation trends.

[0033] The calculation unit 12 calculates, for example, the wiper operation ratio, which indicates the operating status of the wiper system. The wiper operation ratio includes feature quantities H, I, and J calculated from the maximum number of vehicles Mn whose wiper system was operated in each predetermined period (for example, 10 minutes), and feature quantities K, L, and M calculated from the sum of the number of vehicles Mn whose wiper system was operated in each predetermined period (for example, 60 minutes).

[0034] For example, if all vehicles Mn operated their wipers at a speed (e.g., 50 times / minute) for 10% of their driving time, the feature value for the percentage of wiper operation time would be calculated as 10%. In this case, since all vehicles Mn had their wipers activated for 60 minutes, the feature value for the percentage of vehicles with wipers activated would be calculated as 100%.

[0035] In calculating features, calculating the percentage of wiper operation rather than the number of times the wiper device MN operates can suppress variations in calculated values ​​caused by differences in the number of vehicles Mn within the calculation area. Calculating features related to the percentage of wiper operation time can make it easier to capture heavy rainfall. In calculating features, using the "number of vehicles" perspective in addition to the "time" perspective can make it easier to capture heavy rainfall.

[0036] The calculation unit 12 generates a second feature quantity Tb (feature quantity N~Y) based on second detection data obtained from a second region Rb adjacent to the first region Ra included in the observation target region RX. The second region Rb is defined by multiple data collection regions, for example, which include surrounding points adjacent to the first region Ra. The multiple data collection regions are defined by a 1km square rectangular region, for example, which includes surrounding points located at a predetermined distance from the user's position. The surrounding points include, for example, a point 10km away from the user's position in a predetermined direction, a point 20km away from the user's position, and a point 30km away from the user's position in a predetermined direction.

[0037] The above features indicating precipitation levels are calculated for regions 10Rm, 20Rm, and 30Rm, which are separated in four directions around region Rm within the second region Rb. The wider the sampling area used as the basis for calculating the features according to direction, the more vehicles Mn can collect data. Furthermore, by sampling region Rm according to direction, it is possible to capture precipitation trends.

[0038] The calculation unit 12 collects, for example, second detection data of the second region Rb from the user's position. The calculation unit 12 acquires second detection data of the second region Rb in a second predetermined period t2 that is past the first predetermined period t1. The second predetermined period t2 is, for example, one hour past the start time of the first predetermined period t1. The calculation unit 12 generates feature quantities N to Y in the four cardinal directions, for example, based on the second detection data acquired in the second region Rb. The calculation unit 12 generates feature quantities N to Y based on wiper operation data that indicates the operating status of a wiper device installed on a vehicle, for example.

[0039] The calculation unit 12 calculates, for example, the wiper operating time ratio, which indicates the operating status of the wiper device MN. The wiper operating time ratio includes feature quantities N, O, P calculated from the maximum operating time of the wiper device MN in each predetermined period (e.g., 10 minutes), and feature quantities Q, R, S calculated from the sum of the operating times of the wiper device MN in a predetermined period (e.g., 60 minutes).

[0040] Feature quantities N and Q are calculated within a 1km x 1km rectangular area 10Rm (see Figure 2) that includes a point 10km away from the user's location. Feature quantities O and R are calculated within a 1km x 1km rectangular area 20Rm (see Figure 2) that includes a point 20km away from the user's location. Feature quantities P and S are calculated within a 1km x 1km rectangular area 30Rm (see Figure 2) that includes a point 30km away from the user's location.

[0041] The calculation unit 12 calculates, for example, the wiper operation rate, which indicates the operating status of the wiper device MN. The wiper operation rate includes feature quantities T, U, V calculated from the maximum number of vehicles Mn on which the wiper device was operated during a predetermined period (e.g., 10 minutes), and feature quantities W, X, Y calculated from the sum of the number of vehicles Mn on which the wiper device was operated during a predetermined period (e.g., 60 minutes). The calculation unit 12 calculates feature quantities including the first feature quantity Ta and the second feature quantity Tb in four directions from the user's position.

[0042] As described above, the first feature quantity Ta in the first region Ra, which includes the user's position, is calculated in the most recent first predetermined period t1. The second feature quantity Tb in the second region Rb, which is farther from the user's position, is calculated in the second predetermined period t2, which is further back than the first predetermined period t1. The calculation unit 12 inputs the calculated first and second feature quantities into a pre-set precipitation estimation model and calculates the current and future precipitation in the first region Ra, which is included in the observation target region RX.

[0043] The precipitation estimation model is set up using, for example, a polynomial nonlinear regression equation that calculates an estimated precipitation value for a region Rm including the user's location, using multiple features as variables. The nonlinear regression equation can be arbitrarily set to a function capable of calculating precipitation. The relationship between precipitation and features related to the operating status of the wiper device MN gradually changes from a linear relationship to a curvilinear relationship. Therefore, a precipitation estimation model using nonlinear regression can more accurately reproduce the different trends between heavy and light precipitation levels.

[0044] In the precipitation estimation model, multiple parameters corresponding to a polynomial consisting of multiple features are set in the calculation unit 12 based on machine learning performed in advance. The calculation unit 12 repeatedly performs machine learning, for example, using deep learning with a neural network.

[0045] The calculation unit 12 adjusts the parameters set in the precipitation estimation model by repeatedly performing machine learning, which uses features calculated based on past detection data as training data and calculates estimated precipitation values ​​based on these features. By inputting the features calculated using the detection data into the precipitation estimation model with adjusted parameters, an estimated precipitation value is calculated. The precipitation estimation model with adjusted parameters is stored as a computer program in the storage unit 14. The calculation unit 12 calculates the precipitation at any given location by inputting the features calculated using the detection data into the precipitation estimation model.

[0046] The calculation unit 12 acquires detection data related to precipitation detected by multiple vehicles Mn located in the observation area RX. Based on the detection data, the calculation unit 12 calculates the amount of precipitation from the present to the future in a predetermined area included in the observation area RX.

[0047] The calculation unit 12 calculates a plurality of first feature quantities Ta for a first predetermined period t1 in the past compared to the present, based on first detection data acquired from the first region Ra. The calculation unit 12 calculates a second feature quantity Tb for a second predetermined period t2 in the second region Rb in the past compared to the first predetermined period, based on second detection data acquired from the second region Rb adjacent to the first region Ra. The calculation unit 12 calculates a feature quantity that includes a first feature quantity Ta relating to the proportion of time the wiper device MN was in operation and a second feature quantity Tb relating to the proportion of the number of wiper devices that were in operation, based on wiper operation data from the detection data that indicates the operating status of the wiper device MN installed on the vehicle Mn.

[0048] The calculation unit 12 calculates the current precipitation in region Rm using a precipitation estimation model with the first feature quantity Ta and the second feature quantity Tb as variables. The calculation unit 12 compares the calculated precipitation value with a precipitation level set according to the precipitation to calculate the precipitation level in region Rm. For example, the calculation unit 12 determines one of the following three precipitation levels based on the comparison result with a pre-set standard precipitation amount.

[0049] The calculation unit 12 determines that the precipitation level is "High" if the amount of precipitation is equal to or greater than a preset first precipitation amount V1 (for example, 20 mm / h). The High Precipitation Level is set to an amount of precipitation equal to or greater than a threshold requiring attention (for example, 20 mm / h). The threshold can be set to any amount of precipitation. The calculation unit 12 determines that the precipitation level is "Mid" if the amount of precipitation is less than the first precipitation amount V1 and equal to or greater than a preset second precipitation amount V2 (for example, 1 mm / h). The calculation unit 12 determines that there is "No Precipitation (Low)" if the amount of precipitation is less than the second precipitation amount V2 (1 mm / h).

[0050] The calculation unit 12 may calculate not only the precipitation level in the region Rm including the user's position, but also the precipitation levels in all other regions Rm within the observation target region RX by repeatedly performing the same calculation process as described above. The calculation unit 12 does not necessarily have to calculate all other regions Rm within the observation target region RX; it may calculate the observation target region RX using one region Rm contained within a rectangular region of arbitrary size as a representative value, and generate mapping data for the observation target region RX divided by rectangular regions of arbitrary size.

[0051] As shown in Figure 5, the calculation unit 12 may generate mapping data PP that maps precipitation levels for each region Rm in the observation target region RX. As shown in the figure, the mapping data PP shows the distribution of regions with current heavy precipitation levels (High), regions with light precipitation levels (Mid), and regions with no precipitation (Low).

[0052] The precipitation estimation model may be configured to calculate not only current precipitation but also future precipitation by repeatedly performing machine learning based on past detection data. Accordingly, the calculation unit 12 may calculate past and future precipitation levels in region Rm based on the precipitation estimation model. The calculation unit 12 may, for example, calculate precipitation a predetermined time ago, current precipitation, and precipitation after a predetermined time in region Rm including the user's location. The calculation unit 12 may calculate past, present, and future precipitation at any time interval. The calculation unit 12 may update the precipitation level estimation model at predetermined timings using accumulated detection data.

[0053] Generally, heavy precipitation events often occur as large masses moving from distant locations, linked to the movement of rain clouds. Therefore, by calculating the first feature, Ta, in the first predetermined period t1, and the second feature, Tb, in the second predetermined period t2, the tendency of precipitation movement can be calculated.

[0054] The calculation unit 12 may extract areas Rm within the observation target area RX where the precipitation amount is above a certain threshold, based on the generated mapping data, and provide information about the precipitation to users in area Rm. The calculation unit 12 may also cause the notification unit ML of the vehicle Mn in which the extracted users in area Rm are riding to notify the notification unit ML of a certain precipitation level via the network NW. The calculation unit 12 may also extract areas Rm where the precipitation level will be high within a predetermined time period in the future, based on the generated mapping data, and provide information about the precipitation to users in area Rm. The predetermined time period in the future may be set arbitrarily.

[0055] Vehicle Mn may perform driving control in response to rainfall based on predetermined information provided. Driving control in response to rainfall may be applied to vehicle Mn that is driving based on autonomous driving. Driving control in response to rainfall may also be used to assist vehicle Mn that is driving based on manual driving. The calculation unit 12 may cause the display unit 26 of the user's terminal device 20 to display predetermined information regarding the first precipitation level via the network NW.

[0056] The display unit 26 displays, for example, information regarding the state of heavy rainfall or information indicating that heavy rainfall is approaching. The user's terminal device 20 does not necessarily have to be located within the extracted region Rm. The user only needs to be a user of a service that provides rainfall level information.

[0057] Figure 7 shows a flowchart illustrating the flow of the precipitation level estimation method performed in the precipitation level estimation system 1. The precipitation level estimation method is processed based on a program installed on the computer that constitutes the precipitation level estimation system 1. The program causes the computer to execute the following steps related to the precipitation level estimation method.

[0058] The calculation unit 12 acquires detection data related to precipitation detected by multiple vehicles in the observation target area (step S100). Based on the first detection data acquired from the first area Ra included in the observation target area RX, the calculation unit 12 calculates a first feature quantity for a first predetermined period in the past compared to the present (step S102). Based on the second detection data acquired in the second area Rb adjacent to the first area Ra, the calculation unit 12 calculates a second feature quantity for a second predetermined period t2 in the second area Rb compared to a first predetermined period t1 (step S104).

[0059] The calculation unit 12 calculates the amount of precipitation in a predetermined region Rm included in the observation target region RX using a precipitation estimation model with the first feature quantity Ta and the second feature quantity Tb as variables (step S106). The calculation unit 12 determines whether the amount of precipitation is equal to or greater than the first precipitation amount V1 (step S108). If the amount of precipitation is equal to or greater than the first precipitation amount V1, the calculation unit 12 determines that it is a "strong precipitation level" (step S110).

[0060] The calculation unit 12 determines whether the precipitation is equal to or greater than the second precipitation amount V2 if the precipitation is less than the first precipitation amount V1 (step S112). If the precipitation is equal to or greater than the second precipitation amount V2, the calculation unit 12 determines that it is a "weak precipitation level" (step S114). If the precipitation is less than the second precipitation amount V2, the calculation unit 12 determines that there is "no precipitation" (step S116). The calculation unit 12 provides information about the precipitation level to users located in areas with a strong precipitation level (step S118).

[0061] As described above, the precipitation level estimation system 1 can accurately estimate precipitation levels by calculating multiple features based on detection data regarding the operating status of the wiper device MN. The precipitation level estimation system 1 can accurately estimate precipitation levels by inputting multiple features into a precipitation amount estimation model whose parameters have been adjusted based on machine learning. The precipitation level estimation system 1 can accurately estimate precipitation levels by using nonlinear regression in the precipitation amount estimation model.

[0062] In the embodiment described above, the feature quantities were calculated based on data of the operating status of the wiper device MN. The feature quantities may be calculated not only based on the detection data of the wiper device MN, but also based on the detection data of the rain sensor ME, the image data of the surrounding environment of the vehicle Mn captured by the camera MC, and the detection data detected by the lidar device MD. The feature quantities may also be calculated by the calculation unit 12 repeatedly performing machine learning based on the image data of the rainfall situation captured by the camera MC.

[0063] The feature vectors may be calculated by the calculation unit 12 repeatedly performing machine learning based on rainfall detection data detected by the rain sensor ME. Since the detection data of the lidar device MD deteriorates due to rainfall, the feature vectors may be calculated by the calculation unit 12 repeatedly performing machine learning based on the detection data of the lidar device MD. The feature vectors may be calculated using one or more combinations of the detection data of the wiper device MN, camera MC, rain sensor ME, and lidar device MD. The feature vectors may be calculated using any detection data related to precipitation that can be detected.

[0064] The computer programs executed in each configuration of the precipitation level estimation system 1 may be provided in the form of being recorded on a computer-readable, portable recording medium such as semiconductor memory, magnetic recording medium, or optical recording medium. [Explanation of Symbols]

[0065] 1 Precipitation level estimation system, 10 Server device, 12 Calculation unit, 14 Storage unit, 16 Communication unit, 20 Terminal device, 22 Control unit, 24 Storage unit, 26 Display unit, 28 Communication unit, MC Camera, MD LiDAR device, ME Rain sensor, MG Position sensor, ML Notification unit, Mn Vehicle, MN Wiper device, MP Control unit, MS Detection unit, MU Communication unit, NW Network, PP Mapping data, Ra First region, Rb Second region, Rm Region, RX Observation target region

Claims

1. We acquire precipitation-related detection data detected by multiple vehicles located within the observation area. The system includes a calculation unit that calculates the amount of precipitation in a predetermined area included in the observation target area based on the detection data, The aforementioned arithmetic unit, Based on the first detection data obtained from the first region including the predetermined region, a first feature quantity is calculated for the first predetermined period in the past compared to the present. Based on the second detection data acquired in the second region adjacent to the first region, a second feature quantity is calculated for the second predetermined period in the second region that is past the first predetermined period. The precipitation is calculated using a precipitation estimation model with the first and second features as variables, and the precipitation level corresponding to the precipitation is calculated. Precipitation level estimation system.

2. The aforementioned arithmetic unit, Based on the wiper operation data from the detection data that indicates the operating status of the wiper device installed on the vehicle, a feature quantity is calculated that includes the first feature quantity relating to the proportion of time the wiper device was in operation and the second feature quantity relating to the proportion of the number of vehicles in which the wiper device was in operation, and the amount of precipitation is calculated based on the feature quantity. A precipitation level estimation system according to claim 1.

3. The calculation unit calculates the precipitation level in the region and generates mapping data that maps the precipitation level to each of the multiple regions in the observation target region. A precipitation level estimation system according to claim 1.

4. The calculation unit extracts the regions within the observation target region where the amount of precipitation is above a certain threshold, based on the mapping data, and provides information regarding the precipitation level to users located in those regions. The precipitation level estimation system according to claim 3.

5. We acquire precipitation-related detection data detected by multiple vehicles located within the observation area. Based on the first detection data obtained from the first region included in the aforementioned observation target region, a first feature quantity is calculated for a first predetermined period in the past compared to the present. Based on the second detection data acquired in the second region adjacent to the first region, a second feature quantity is calculated for the second predetermined period in the second region that is past the first predetermined period. A program that causes a computer to perform a process of calculating the amount of precipitation in a predetermined area included in the observation target area using a precipitation estimation model with the first feature quantity and the second feature quantity as variables, and calculating a precipitation level corresponding to the amount of precipitation.

Images