JP7616119B2 - Information processing device, information processing method, and information processing program - Google Patents

Description

The present disclosure relates to an information processing device, an information processing method, and an information processing program.

Conventionally, a road surface evaluation device that eliminates sensor installation location dependency in a road surface diagnosis method using vehicle response is known (see, for example, Patent Document 1). The road surface evaluation device of Patent Document 1 synchronizes and records the pitching angular velocity of the vehicle acquired by an angular velocity sensor with GPS information acquired by a GPS. Then, the road surface evaluation device estimates the acceleration response of a quarter car, which is a virtual vehicle that serves as a reference, from the angular velocity response of the measurement vehicle using a transfer function from the acquired pitching angular velocity of the vehicle. Then, the road surface evaluation device estimates the International Roughness Index (IRI) from the estimated acceleration response of the quarter car using a correlation function. Note that the technology of Patent Document 1 proposes an IRI estimation method using a smartphone (for example, paragraph [0051]), in which the smartphone is fixed to the front center of the vehicle using double-sided tape (for example, paragraph [0054]), and various values are measured using the smartphone.

There is also known a technology that allows users to more easily plan maintenance plans according to the type of road abnormality (see, for example, Patent Document 2). The display processing device of Patent Document 2 detects a first abnormal section, which is a road section whose road condition is a first abnormality, and a second abnormal section, which is a road section whose road condition is a second abnormality different from the first abnormality, based on vehicle information from multiple vehicles. Then, the display processing device assigns status information in a first display mode to the first abnormal section among the roads on the displayed map, and assigns status information in a second display mode different from the first display mode to the second abnormal section, and displays them on the display device.

JP 2015-028456 A Patent Publication No. 2021-033130

The technology disclosed in Patent Document 1 above cannot obtain road surface condition information (e.g., road surface roughness information that indicates the roughness of the road surface) that indicates the condition of the road surface on which a vehicle that is not equipped with a smartphone has traveled. For this reason, the technology in Patent Document 1 above has the problem that it cannot obtain road surface condition information for a larger number of roads.

The display processing device disclosed in Patent Document 2 estimates road conditions based on vehicle information from multiple vehicles and displays the road conditions. However, the display processing device disclosed in Patent Document 2 requires complex processing, such as determining that the road conditions in the target section are abnormal when the maximum wheel speed fluctuation rate in the target section for all target vehicles is equal to or greater than a threshold value, or when the avoidance behavior ratio is equal to or greater than a threshold value (for example, paragraph [0093] of Patent Document 2). For this reason, the technology in Patent Document 2 has the problem that it cannot easily estimate road surface condition information.

As a result, conventional technology has the problem that it is not possible to easily estimate road surface conditions from vehicle condition information of multiple vehicles.

Taking the above facts into consideration, the present disclosure aims to easily estimate road surface conditions from vehicle condition information of multiple vehicles using a trained model generated in advance by machine learning.

A first aspect of the present disclosure is an information processing device that includes an acquisition unit that acquires vehicle state information for each of a plurality of vehicles, and an estimation unit that estimates road surface state information on which each of the plurality of vehicles has traveled based on the vehicle state information for each of the plurality of vehicles acquired by the acquisition unit, and the estimation unit is a trained model that outputs the road surface state information when the vehicle state information is input, and estimates the road surface state information by inputting the vehicle state information acquired by the acquisition unit to a trained model that has been trained in advance based on training data in which the vehicle state information and the road surface state information are associated with each other.

The information processing device of the first aspect of the present disclosure acquires vehicle state information for each of a plurality of vehicles. The information processing device estimates road surface state information on which each of the plurality of vehicles has traveled, based on the acquired vehicle state information for each of the plurality of vehicles. The information processing device estimates road surface state information by inputting vehicle state information to a trained model that outputs road surface state information when vehicle state information is input, and that has been trained in advance based on learning data in which the vehicle state information and the road surface state information are associated with each other. According to the information processing device of the first aspect of the present disclosure, it is possible to easily estimate road surface states from the vehicle state information of a plurality of vehicles, using a trained model that has been generated in advance by machine learning.

The information processing device of the second aspect of the present disclosure further includes a driving route generation unit that generates a vehicle driving route without including a road corresponding to the road surface in the vehicle driving route when the road surface condition information of the road surface indicates an abnormal state. This makes it possible to present a driving route that avoids roads with abnormal road surfaces.

The vehicle condition information of the information processing device of the third aspect of the present disclosure includes at least one of information representing the longitudinal vehicle body acceleration of the vehicle, information representing the lateral vehicle body acceleration of the vehicle, and information representing the yaw rate of the vehicle. This makes it possible to accurately estimate road surface condition information using at least one of information representing the longitudinal vehicle body acceleration of the vehicle, information representing the lateral vehicle body acceleration of the vehicle, and information representing the yaw rate of the vehicle that is considered to be affected by road surface conditions.

The vehicle condition information of the information processing device of the fourth aspect of the present disclosure includes wheel speed information representing the wheel speed of the vehicle. This makes it possible to accurately estimate road surface condition information using the wheel speed that is considered to be affected by the road surface condition.

The estimation unit of the information processing device of the fifth aspect of the present disclosure calculates a differential value of the wheel speed of the vehicle by differentiating wheel speed information representing the wheel speed of the vehicle, which is included in the vehicle condition information of the vehicle, and estimates the road surface condition information of the road surface on which the vehicle has traveled by inputting the differential value of the wheel speed of the vehicle to the trained model. This makes it possible to accurately estimate road surface condition information using the differential value of the wheel speed that is thought to be affected by the road surface condition.

The acquisition unit of the information processing device of the sixth aspect of the present disclosure acquires the driving information from a storage unit in which driving information representing a combination of the vehicle state information for each of a plurality of vehicles and position information at the time when the vehicle state information was acquired is stored, and the estimation unit estimates the road surface state information on which the vehicle traveled for each of the plurality of vehicles based on the vehicle state information in the driving information of the vehicle acquired by the acquisition unit, and further includes a generation unit that generates map data to which the road surface state information has been added by adding the road surface state information to a position in map data corresponding to the position information based on the position information in the driving information of the vehicle acquired by the acquisition unit and the estimated road surface state information for each of the plurality of vehicles. This makes it possible to visualize information regarding the road surface state of a plurality of roads.

The information processing method of the seventh aspect of the present disclosure is an information processing method in which a computer executes a process to acquire vehicle state information for each of a plurality of vehicles, estimate road surface state information on which each of the plurality of vehicles has traveled based on the acquired vehicle state information for each of the plurality of vehicles, and estimate the road surface state information by inputting the acquired vehicle state information into a trained model that outputs the road surface state information when the vehicle state information is input when estimating the road surface state information, and that has been trained in advance based on learning data in which the vehicle state information and the road surface state information are associated with each other. As with the first aspect, the seventh aspect of the present disclosure is an information processing method in which a computer executes a process to acquire vehicle state information for each of a plurality of vehicles, estimates the road surface state information by inputting the acquired vehicle state information into a trained model that is trained in advance based on learning data in which the vehicle state information and the road surface state information are associated with each other. As with the first aspect, the seventh aspect of the present disclosure can easily estimate road surface states from the vehicle state information of a plurality of vehicles using a trained model that is generated in advance by machine learning.

The information processing program of the eighth aspect of the present disclosure is an information processing program for causing a computer to execute a process of acquiring vehicle state information for each of a plurality of vehicles, estimating road surface state information on which each of the plurality of vehicles has traveled based on the acquired vehicle state information for each of the plurality of vehicles, and estimating the road surface state information by inputting the acquired vehicle state information into a trained model that outputs the road surface state information when the vehicle state information is input, and that has been trained in advance based on learning data in which the vehicle state information and the road surface state information are associated with each other. As with the first aspect, the eighth aspect of the present disclosure is an information processing program for causing a computer to execute a process of acquiring vehicle state information for each of a plurality of vehicles, estimating road surface state information from the vehicle state information of the plurality of vehicles, using a trained model that has been generated in advance by machine learning.

A ninth aspect of the present disclosure is an information processing device including: an acquisition unit that acquires driving information from a storage unit that stores driving information representing a combination of vehicle state information representing a driving state of the vehicle obtained when the vehicle is driven and position information on the road where the vehicle has driven; an estimation unit that estimates road surface roughness information of a road surface on which the vehicle has driven based on the vehicle state information in the driving information of the vehicle acquired by the acquisition unit for each of the multiple vehicles; and a generation unit that generates map data to which the road surface roughness information has been added by adding the road surface roughness information to a position in map data corresponding to the position information for each of the multiple vehicles based on the position information in the driving information of the vehicle acquired by the acquisition unit and the road surface roughness information estimated by the estimation unit for each of the multiple vehicles.

The information processing device of the ninth aspect of the present disclosure acquires driving information for each of a plurality of vehicles from a storage unit in which driving information representing a combination of vehicle state information representing the driving state of the vehicle obtained when the vehicle is traveling and position information on the road on which the vehicle has traveled is stored. Then, for each of the plurality of vehicles, the information processing device estimates road surface roughness information of the road surface on which the vehicle has traveled based on the vehicle state information in the acquired vehicle driving information. For each of the plurality of vehicles, the information processing device generates map data to which road surface roughness information has been added by adding road surface roughness information to a position in the map data corresponding to the position information based on the position information in the acquired vehicle driving information and the estimated road surface roughness information. This makes it possible to visualize information regarding the road surface roughness of a plurality of roads.

The vehicle condition information of the information processing device of the tenth aspect of the present disclosure includes wheel speed information representing the wheel speed of the vehicle. This makes it possible to accurately estimate road surface roughness information using the wheel speed that is considered to be affected by road surface roughness.

The estimation unit of the information processing device of the eleventh aspect of the present disclosure is a trained model that outputs road surface roughness information on which the vehicle has traveled when the vehicle state information of the vehicle is input, and estimates road surface roughness information of the road surface on which the vehicle has traveled by inputting the vehicle state information acquired by the acquisition unit into the trained model that has been trained in advance based on training data in which the vehicle state information for training is associated with the road surface roughness information for training. This makes it possible to accurately estimate road surface roughness information using a trained model generated based on training data in which vehicle state information for training is associated with road surface roughness information for training.

The estimation unit of the information processing device of the twelfth aspect of the present disclosure calculates a differential value of the wheel speed of the vehicle by differentiating wheel speed information representing the wheel speed of the vehicle, which is included in the vehicle state information of the vehicle, and estimates road surface roughness information of the road surface on which the vehicle has traveled by inputting the vehicle state information including the differential value of the wheel speed of the vehicle to the trained model. This makes it possible to accurately estimate road surface roughness information using the differential value of the wheel speed that is thought to be affected by road surface roughness.

The vehicle state information of the information processing device of the thirteenth aspect of the present disclosure includes at least one of information representing the longitudinal vehicle body acceleration of the vehicle, information representing the lateral vehicle body acceleration of the vehicle, and information representing the yaw rate of the vehicle. This makes it possible to accurately estimate road surface roughness information using at least one of information representing the longitudinal vehicle body acceleration of the vehicle, information representing the lateral vehicle body acceleration of the vehicle, and information representing the yaw rate of the vehicle that is considered to be affected by road surface roughness.

A fourteenth aspect of the present disclosure is an information processing method in which a computer executes a process to acquire driving information representing a combination of vehicle state information representing a driving state of a vehicle obtained when the vehicle is traveling and position information on a road where the vehicle has traveled from a storage unit in which the driving information is stored, estimate road surface roughness information of a road surface on which the vehicle has traveled based on the vehicle state information in the acquired driving information of the vehicle for each of the plurality of vehicles, and generate map data to which the road surface roughness information has been added by adding the road surface roughness information to a position in map data corresponding to the position information based on the position information in the acquired driving information of the vehicle and the estimated road surface roughness information for each of the plurality of vehicles. The fourteenth aspect of the present disclosure is capable of visualizing information on the road surface roughness of a plurality of roads, similar to the ninth aspect.

A fifteenth aspect of the present disclosure is an information processing program for causing a computer to execute a process of acquiring driving information representing a combination of vehicle state information representing the driving state of a vehicle obtained when the vehicle is traveling and position information on the road on which the vehicle has traveled from a storage unit storing the driving information, estimating road surface roughness information of the road surface on which the vehicle has traveled based on the vehicle state information in the acquired driving information of the vehicle for each of the plurality of vehicles, and generating map data to which the road surface roughness information has been added by adding the road surface roughness information to a position in map data corresponding to the position information based on the position information in the acquired driving information of the vehicle and the estimated road surface roughness information for each of the plurality of vehicles. The fifteenth aspect of the present disclosure is capable of visualizing information on the road surface roughness of a plurality of roads, similar to the ninth aspect.

As described above, the present disclosure has the advantage of being able to easily estimate road surface conditions from vehicle condition information of multiple vehicles using a trained model that has been generated in advance through machine learning.

1 is a schematic block diagram of an information processing system according to an embodiment. 4A to 4C are diagrams for explaining an example of travel information and sensor information according to an embodiment; FIG. 1 is a diagram for explaining a trained model of an embodiment. FIG. 13 is a diagram showing a single wheel spring mass model. FIG. 4 is a diagram showing the relationship between tire radius and ground load. FIG. 1 is a diagram illustrating an example of learning data used to generate a trained model. FIG. 2 is a diagram showing an example of map data with IRIs of road surfaces; FIG. 2 is a diagram illustrating an example of the configuration of a computer that constitutes an information processing device. FIG. 2 is a diagram illustrating an example of information processing executed by the information processing device according to the embodiment.

<Embodiment>

The information processing system of this embodiment will be described below with reference to the drawings.

Figure 1 is a block diagram showing an example of the configuration of an information processing system 10 according to this embodiment. As shown in Figure 1, the information processing system 10 includes multiple vehicles 11-1, 11-2, and 11-3, and an information processing device 13. The multiple vehicles 11-1, 11-2, and 11-3 are communicatively connected to the information processing device 13 via a predetermined communication means NW. In the following, any one of the multiple vehicles 11-1, 11-2, and 11-3 will simply be referred to as "vehicle 11." In the example of Figure 1, three vehicles 11-1, 11-2, and 11-3 are shown, but more vehicles may be present.

(Vehicle 11)
The vehicle 11 is equipped with a GPS (Global Positioning System) 110, a wheel speed sensor 112, an acceleration sensor 114, a yaw rate sensor 116, and a vehicle control device 118. The vehicle control device 118 is, for example, an ECU (Electronic Control Unit).

The GPS 110 sequentially acquires the location information of the vehicle 11.

The wheel speed sensor 112 sequentially acquires wheel speed information representing the wheel speed of the vehicle 11.

The acceleration sensor 114 sequentially acquires front-rear vehicle body acceleration information representing the front-rear vehicle body acceleration of the vehicle 11, and left-right vehicle body acceleration information representing the left-right vehicle body acceleration of the vehicle 11.

The yaw rate sensor 116 sequentially acquires yaw rate information that indicates the yaw rate of the vehicle 11.

(Vehicle control device 118)
As shown in FIG. 1, the vehicle control device 118 functionally comprises a control unit 120, an information storage unit 122, and a transmission/reception unit 124.

The control unit 120 sequentially acquires the position information acquired by the GPS 110, the wheel speed information acquired by the wheel speed sensor 112, the front-rear vehicle body acceleration information and the left-right vehicle body acceleration information acquired by the acceleration sensor 114, and the yaw rate information acquired by the yaw rate sensor 116, and stores this information in the information storage unit 122 in association with time information. The time information indicates the time when each piece of information was acquired by each sensor.

The information storage unit 122 stores the vehicle 11's position information, wheel speed information, front and rear vehicle body acceleration information, left and right vehicle body acceleration information, and yaw rate information at each time.

The transmitter/receiver 124 transmits the vehicle 11 position information, wheel speed information, front/rear vehicle body acceleration information, left/right vehicle body acceleration information, and yaw rate information stored in the information storage unit 122, as well as the time information when the information was acquired, to the information processing device 13 every time a predetermined time has elapsed. Note that hereinafter, the wheel speed information, front/rear vehicle body acceleration information, left/right vehicle body acceleration information, and yaw rate information are also simply referred to as "sensor information".

(Information processing device 13)
As shown in FIG. 1 , the information processing device 13 functionally includes a reception unit 130 , a driving information storage unit 132 , a learned model storage unit 134 , an acquisition unit 136 , an estimation unit 138 , a generation unit 140 , and a display control unit 142 .

The reception unit 130 receives various information transmitted from the vehicle control device 118 of each of the multiple vehicles 11-1, 11-2, and 11-3. Specifically, the reception unit 130 receives a combination of position information, sensor information, and time information transmitted from the vehicle control device 118 of each of the multiple vehicles 11-1, 11-2, and 11-3, and stores this information as driving information in the driving information storage unit 132.

The driving information storage unit 132 stores driving information, which is a combination of position information and sensor information transmitted from each of the vehicle control devices 118 of the multiple vehicles 11-1, 11-2, and 11-3.

Figure 2 shows an example of driving information stored in the driving information storage unit 132. In the example shown in Figure 2, a vehicle ID, which is identification information for the vehicle 11, time information, position information, and sensor information are stored in association with each other. The combination of position information and sensor information at each time transmitted from the vehicle control device 118 constitutes the driving information.

The trained model storage unit 134 stores a trained model that outputs road surface condition information on which the vehicle 11 has traveled when differential value information of the wheel speed of the vehicle 11, front-rear vehicle body acceleration information, left-right vehicle body acceleration information, and yaw rate information and time information are input. In the following, the wheel speed differential value information, front-rear vehicle body acceleration information, left-right vehicle body acceleration information, and yaw rate information are also simply referred to as "vehicle condition information". The trained model of this embodiment outputs road surface roughness information that indicates the roughness of the road surface as an example of road surface condition information.

FIG. 3 shows an example of the trained model of this embodiment. As shown in FIG. 3, when time information and vehicle state information are input, the trained model of this embodiment outputs road surface roughness information of the road surface on which the vehicle 11 has traveled. In addition, in the wheel speed differential value information, "FR" represents the differential value of the wheel speed of the right front wheel of the vehicle 11, "FL" represents the differential value of the wheel speed of the left front wheel of the vehicle 11, "RR" represents the differential value of the wheel speed of the right rear wheel of the vehicle 11, and "RL" represents the differential value of the wheel speed of the left rear wheel of the vehicle 11. Since the trained model of this embodiment is a LSTM (Long Short Term Memory) network, which is a type of neural network model, time information when the vehicle state information is acquired is also input. In addition, a, b, c, ..., h of the trained model in FIG. 3 represent each node of the input layer, and A, B, C, D, ... represent each node of the intermediate layer. In the example of Figure 3, there are eight nodes in the input layer and 200 nodes in the intermediate layer. Weight parameters are added to the edges connecting these nodes, and these weight parameters are adjusted in advance by machine learning.

As shown in FIG. 3, the trained model of this embodiment outputs the International Roughness Index (IRI), which is an example of road surface roughness information. Using this trained model, the IRI of the road surface on which the vehicle 11 has traveled is estimated. In this embodiment, the differential value of the wheel speed is used as a factor for estimating the IRI of the road surface.

Here, we explain the relationship between the road surface IRI and the differential value of the wheel speed.

Fig. 4 shows a diagram of a known single wheel spring mass model. As shown in Fig. 4, the single wheel spring mass model includes a sprung mass m _s , an unsprung mass m _u , a coil Co, an absorber Ab, and a simulated tire Ti. Note that X _S in Fig. 4 represents a sprung mass, which represents the displacement of the sprung mass m _s . Also, X _U in Fig. 4 represents an unsprung mass, which represents the displacement of the unsprung mass m _u . X _O represents the ground, which represents the displacement of the contact surface.

As shown in the formula below, the IRI of the road surface is the ratio of the cumulative vertical motion displacement experienced by the vehicle when it is driven at a constant speed in a single wheel spring mass model to the travel distance L. Note that the dots in the formula below indicate differentiation.

In this embodiment, the IRI of the road surface is estimated using a trained model without calculating the above formula.

Fig. 5 shows the relationship between tire radius and ground contact load. In Fig. 5, V represents vehicle speed, r represents tire radius, ω represents wheel speed, O represents the center of tire T, _Ft represents the ground contact load received by tire T, and δ represents the amount of change.

Considering the single wheel spring mass model shown in Fig. 4 and the relationship shown in Fig. 5, the following equation is derived: _Kt = Kt + ...

The above formula is established between the change amount δω of the wheel speed and the change amount _δFt of the ground load applied to the tire. Therefore, it can be seen that the wheel speed ω is affected by the ground load _Ft applied to the tire. As shown in FIG. 5, the ground load _Ft applied to the tire T is an amount related to the IRI, which is the road surface roughness. Therefore, it is considered that the IRI of the road surface can be estimated with high accuracy by using the wheel speed ω when estimating the IRI of the road surface.

Therefore, in this embodiment, the IRI of the road surface is estimated by using the wheel speed ω when estimating the IRI of the road surface. Specifically, the IRI of the road surface is estimated with high accuracy by including a differential value of the wheel speed ω in the input of the trained model. As shown in the above formula, the change amount δω of the wheel speed is related to the change amount _δFt of the ground load, so in this embodiment, the IRI of the road surface is estimated by inputting the differential value of the wheel speed ω to the trained model.

The trained model in FIG. 3 is generated by performing machine learning in advance based on training data collected in advance. Specifically, the trained model in this embodiment is generated by performing machine learning in advance based on training data such as that shown in FIG. 6.

The acquisition unit 136 acquires driving information for each of the multiple vehicles 11-1, 11-2, and 11-3 from the driving information storage unit 132.

The estimation unit 138 acquires wheel speed information from the sensor information of the vehicle 11 travel information acquired by the acquisition unit 136 for each of the multiple vehicles 11-1, 11-2, and 11-3. Next, the estimation unit 138 calculates wheel speed differential value information by differentiating the wheel speed information for each of the multiple vehicles 11-1, 11-2, and 11-3. The estimation unit 138 sets a combination of the wheel speed differential value information, front-rear vehicle body acceleration information, left-right vehicle body acceleration information, and yaw rate information as vehicle state information of the vehicle 11 for each of the multiple vehicles 11-1, 11-2, and 11-3.

The estimation unit 138 then inputs the vehicle state information and time information of each of the multiple vehicles 11-1, 11-2, and 11-3 into the trained model stored in the trained model storage unit 134, thereby estimating the IRI of the road surface on which each of the multiple vehicles 11-1, 11-2, and 11-3 has traveled.

The generation unit 140 generates map data with the road surface IRI added to it by adding the road surface IRI to a position in the map data corresponding to the position information of the vehicle 11's driving information for each of the multiple vehicles 11-1, 11-2, 11-3 based on the position information of the vehicle 11 acquired by the acquisition unit 136 and the road surface IRI estimated by the estimation unit 138.

The display control unit 142 controls the map data to which the IRI generated by the generation unit 140 is added to be displayed on a display device (not shown). The display control unit 142 may store the map data to which the IRI of the road surface generated by the generation unit 140 is added in a predetermined storage unit (not shown).

Figure 7 shows an example of map data to which the IRI of the road surface has been added. The dots in Figure 7 represent roads to which an IRI has been added. For example, the map data in Figure 7 may be displayed so that the higher the IRI value of the road surface, the darker the dots are displayed, and the lower the IRI value of the road surface, the lighter the dots are displayed. Alternatively, for example, the IRI value of the road surface may be represented by color. This allows the surface roughness information of multiple roads to be appropriately visualized.

The information processing device 13 can be realized, for example, by a computer 50 as shown in FIG. 8. The computer 50 includes a CPU 51, a memory 52 as a temporary storage area, and a non-volatile storage unit 53. The computer 50 also includes an input/output interface (I/F) 54 to which an input/output device or the like (not shown) is connected, and a read/write (R/W) unit 55 that controls reading and writing of data to a recording medium 59. The computer 50 also includes a network I/F 56 that is connected to a network such as the Internet. The CPU 51, memory 52, storage unit 53, input/output I/F 54, R/W unit 55, and network I/F 56 are connected to one another via a bus 57.

The storage unit 53 can be realized by a hard disk drive (HDD), a solid state drive (SSD), a flash memory, etc. The storage unit 53 as a storage medium stores a program for causing the computer 50 to function. The CPU 51 reads the program from the storage unit 53, expands it into the memory 52, and sequentially executes the processes contained in the program.

Next, the operation of the information processing device 13 of the embodiment will be described.

The vehicle control devices 118 of the multiple vehicles 11-1, 11-2, and 11-3 sequentially transmit position information and sensor information at each time to the information processing device 13, and multiple pieces of driving information are stored in the driving information storage unit 132 of the information processing device 13. Then, when the information processing device 13 receives an instruction signal to generate map data with an IRI, the information processing device 13 executes the information processing routine shown in FIG. 9.

In step S100, the acquisition unit 136 acquires from the driving information storage unit 132 the driving information of each of the multiple vehicles 11-1, 11-2, and 11-3 and time information indicating the time at which the driving information was acquired.

In step S102, the estimation unit 138 reads the trained model stored in the trained model memory unit 134.

In step S104, the estimation unit 138 calculates a differential value of the wheel speed by differentiating the wheel speed information of the vehicle state information of each of the multiple vehicles 11-1, 11-2, and 11-3 acquired in step S100. Then, the estimation unit 138 sets the wheel speed differential value information, the front-rear vehicle body acceleration information, the left-right vehicle body acceleration information, and the yaw rate information as the vehicle state information for each of the multiple vehicles 11-1, 11-2, and 11-3.

In step S106, the estimation unit 138 inputs the time information and vehicle state information into the trained model stored in the trained model storage unit 134, thereby estimating the IRI of the road surface on which each of the multiple vehicles 11-1, 11-2, and 11-3 has traveled.

In step S108, the generation unit 140 generates map data to which the road surface IRI has been added by adding the road surface IRI estimated in step S106 to the position in the map data corresponding to the position information obtained in step S100.

In step S110, the display control unit 142 displays the map data with the IRI generated in step S108 on a display device (not shown), and ends the information processing routine.

As described above, the information processing device 13 according to this embodiment acquires vehicle state information for each of the multiple vehicles. The information processing device 13 estimates road surface state information on which each of the multiple vehicles has traveled, based on the acquired vehicle state information for each of the multiple vehicles. The information processing device 13 estimates road surface state information by inputting vehicle state information to a trained model that outputs road surface state information when vehicle state information is input, and that has been trained in advance based on learning data in which vehicle state information and road surface state information are associated with each other. This makes it possible to easily estimate road surface states from the vehicle state information of the multiple vehicles, using a trained model that has been generated in advance by machine learning.

The information processing device 13 according to this embodiment acquires driving information from a driving information storage unit that stores driving information representing a combination of vehicle state information representing the driving state of the vehicle obtained when the vehicle is driving and position information on the road on which the vehicle has driven for each of the multiple vehicles. The information processing device 13 estimates road surface roughness information of the road surface on which the vehicle has driven for each of the multiple vehicles, based on the vehicle state information in the acquired vehicle driving information. Then, the information processing device 13 generates map data to which road surface roughness information has been added by adding road surface roughness information to a position corresponding to the position information in the map data for each of the multiple vehicles, based on the position information in the acquired vehicle driving information and the estimated road surface roughness information. This makes it possible to visualize information regarding the road surface roughness of multiple roads.

In addition, the information processing device 13 according to this embodiment can accurately estimate the IRI of the road surface by using the vehicle's wheel speed information.

The processing performed by each device in the above embodiment has been described as software processing performed by executing a program, but it may be processing performed by hardware. Alternatively, it may be processing that combines both software and hardware. In addition, the programs stored in the ROM may be stored in various storage media and distributed.

In addition, the processing performed by each device in the above embodiments may be executed by one of various processors, or may be executed by a combination of two or more processors of the same or different types.

Furthermore, the present invention is not limited to the above, and can of course be modified in various ways without departing from the spirit of the invention.

For example, in the above embodiment, a trained model is used to estimate the IRI of the road surface, but this is not limited to the above. For example, a predetermined correlation function such as that disclosed in JP 2015-028456 A may be used.

In addition, in the above embodiment, an example has been described in which an LSTM network is used, but the present invention is not limited to this. For example, other types of neural networks may be used as the trained model. When a neural network other than an LSTM network is used as the trained model, time information as input information may not be necessary.

In the above embodiment, the IRI of the road surface is estimated using the differential value information of the wheel speed, but the present invention is not limited to this. For example, the IRI of the road surface may be estimated using the wheel speed information itself. In this case, the wheel speed information, the front and rear vehicle body acceleration information, the left and right vehicle body acceleration information, and the yaw rate information are set as the vehicle state information, and when the vehicle state information is input to the trained model, the IRI of the road surface is output. In addition, instead of using the differential value information of the wheel speed or the wheel speed information, the front and rear vehicle body acceleration information, the left and right vehicle body acceleration information, and the yaw rate information as the vehicle state information, the IRI of the road surface may be estimated using at least one of these pieces of information as the vehicle state information.

In addition, in the above embodiment, an example was described in which IRI was estimated as road surface roughness information, but this is not limited to this. Other types of road surface roughness information may also be estimated.

In the above embodiment, the road surface roughness information is estimated as the road surface condition information, but the present invention is not limited to this, and other types of road surface conditions may be estimated. For example, foreign object information indicating the degree to which a foreign object exists on the road surface may be estimated as the road surface condition information. In this case, for example, the yaw rate information of the vehicle is input to the trained model, and the trained model outputs a probability indicating the degree to which a foreign object exists as foreign object information. If a foreign object exists on the road surface, it is expected that the vehicle will run in a manner that quickly avoids the foreign object. In this case, the behavior of the vehicle avoiding the foreign object appears in the yaw rate of the vehicle. Therefore, a trained model is generated based on training data in which the yaw rate information for training and the foreign object information for training are associated, and the trained model can be used to easily and accurately estimate foreign object information from the yaw rate information of the vehicle.

Alternatively, for example, slip information indicating the likelihood of a vehicle traveling on a road surface slipping or the likelihood of hydroplaning may be estimated as road surface condition information. For example, in rainy weather and snowfall, vehicles traveling on a road surface are likely to slip. In this case, for example, information indicating at least one of wheel speed, vehicle speed, friction coefficient, and lateral acceleration (e.g., left and right vehicle body acceleration) is input to the trained model, and the trained model outputs a probability indicating the likelihood of a vehicle slipping or the likelihood of hydroplaning as slip information. A trained model is generated based on training data in which information indicating at least one of learning wheel speed, vehicle speed, friction coefficient, and lateral acceleration is associated with learning slip information, and by using the trained model, slip information can be easily and accurately estimated from information indicating at least one of wheel speed, vehicle speed, friction coefficient, and lateral acceleration.

In the above embodiment, the map data with IRI, which is an example of road surface roughness information, is displayed on the display device (not shown), but the present invention is not limited to this. For example, at least one of the information processing device 13 and the vehicle control device 118 may further include a driving route generation unit that generates a driving route for the vehicle. The driving route generation unit has a function of a so-called navigation system. In this case, for example, when the road surface condition information of the road surface indicates an abnormal state, the driving route generation unit of at least one of the information processing device 13 and the vehicle control device 118 may generate a driving route for the vehicle without including the road corresponding to the road surface in the driving route for the vehicle. For example, the driving route generation unit of at least one of the information processing device 13 and the vehicle control device 118 performs a process such as not guiding a road with a bad road surface condition as a driving route based on the estimated road surface condition information. For example, if the road surface roughness information is equal to or greater than a threshold, the foreign object information is equal to or greater than a threshold, or the road surface slippage information is equal to or greater than a threshold, the driving route generation unit determines that the road surface condition is poor or abnormal, and generates a driving route for the vehicle without including the road corresponding to the road surface in the vehicle driving route. This allows a driving route that avoids roads with abnormal road surfaces to be appropriately presented.

REFERENCE SIGNS LIST 10 Information processing system 11 Vehicle 13 Information processing device 50 Computer 112 Wheel speed sensor 114 Acceleration sensor 116 Yaw rate sensor 118 Vehicle control device 120 Control unit 122 Information storage unit 124 Transmitting/receiving unit 130 Reception unit 132 Traveling information storage unit 134 Learned model storage unit 136 Acquisition unit 138 Estimation unit 140 Generation unit 142 Display control unit

Claims (8)

an acquisition unit that acquires vehicle state information for each of a plurality of vehicles;
an estimation unit that estimates road surface condition information on which each of the plurality of vehicles has traveled based on the vehicle condition information for each of the plurality of vehicles acquired by the acquisition unit,
the estimation unit estimates the road surface condition information by inputting the vehicle state information acquired by the acquisition unit to a trained model that outputs the road surface condition information when the vehicle state information is input, the trained model having been trained in advance based on training data in which the vehicle state information and the road surface condition information are associated with each other ;
The trained model is a trained Long Short Term Memory (LSTM) neural network model,
The vehicle state information includes at least one of information representing a longitudinal vehicle body acceleration of the vehicle, information representing a lateral vehicle body acceleration of the vehicle, and information representing a yaw rate of the vehicle, and time information representing a time when the at least one piece of information is acquired by a sensor.
Information processing device. and a travel route generating unit configured to generate a travel route for a vehicle without including a road corresponding to the road surface in the travel route for the vehicle when the road surface condition information indicates an abnormal state.
The information processing device according to claim 1 . The vehicle state information includes wheel speed information representing a wheel speed of the vehicle.
3. The information processing device according to claim 1 or 2 . the estimation unit calculates a differential value of the wheel speed of the vehicle by differentiating wheel speed information representing a wheel speed of the vehicle, which is included in the vehicle state information of the vehicle, and estimates the road surface state information of the road surface on which the vehicle has traveled by inputting the differential value of the wheel speed of the vehicle to the trained model.
The information processing device according to claim 3 . The acquisition unit acquires the travel information from a storage unit in which travel information representing a combination of the vehicle state information for each of a plurality of vehicles and position information at the time when the vehicle state information was acquired is stored,
The estimation unit estimates, for each of a plurality of vehicles, the road surface condition information on which the vehicle has traveled based on the vehicle condition information included in the vehicle travel information acquired by the acquisition unit; and
a generating unit configured to generate map data to which road surface condition information has been added by adding the road surface condition information to a position in map data corresponding to the position information based on the position information of the vehicle in the traveling information of the vehicle acquired by the acquiring unit and the estimated road surface condition information for each of the plurality of vehicles,
The information processing device according to any one of claims 1 to 4 . The road surface condition information is at least one of road surface roughness information indicating the roughness of the road surface, foreign matter information indicating the degree to which foreign matter exists on the road surface, and slippage information indicating the likelihood of a vehicle traveling on the road surface slipping or the likelihood of a hydroplaning phenomenon occurring.
The information processing device according to any one of claims 1 to 5. Obtaining vehicle status information for each of a plurality of vehicles;
estimating road surface condition information on which each of the plurality of vehicles has traveled based on the acquired vehicle condition information for each of the plurality of vehicles;
a trained model that outputs the road surface condition information when the vehicle condition information is input, and that is trained in advance based on learning data in which the vehicle condition information and the road surface condition information are associated with each other, when estimating the road surface condition information, by inputting the acquired vehicle condition information to the trained model;
An information processing method in which a process is executed by a computer, comprising the steps of:
The trained model is a trained Long Short Term Memory (LSTM) neural network model,
The vehicle state information includes at least one of information representing a longitudinal vehicle body acceleration of the vehicle, information representing a lateral vehicle body acceleration of the vehicle, and information representing a yaw rate of the vehicle, and time information representing a time when the at least one piece of information is acquired by a sensor.
Information processing methods . Obtaining vehicle status information for each of a plurality of vehicles;
estimating road surface condition information on which each of the plurality of vehicles has traveled based on the acquired vehicle condition information for each of the plurality of vehicles;
a trained model that outputs the road surface condition information when the vehicle condition information is input, and that is trained in advance based on learning data in which the vehicle condition information and the road surface condition information are associated with each other, when estimating the road surface condition information, by inputting the acquired vehicle condition information to the trained model;
An information processing program for causing a computer to execute a process,
The trained model is a trained Long Short Term Memory (LSTM) neural network model,
The vehicle state information includes at least one of information representing a longitudinal vehicle body acceleration of the vehicle, information representing a lateral vehicle body acceleration of the vehicle, and information representing a yaw rate of the vehicle, and time information representing a time when the at least one piece of information is acquired by a sensor.
Information processing program .