JP7519619B2 - Information processing device, information processing method, and program - Google Patents

Description

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

Hundreds of lightning strikes occur annually during domestic aircraft operations in Japan. Although the likelihood of an aircraft being struck by lightning directly leading to a serious accident is extremely low, it is estimated that repairing damage to the exterior of the aircraft costs several times that amount, amounting to hundreds of millions of yen, annually. In addition, since inspections and emergency treatment of aircraft that have been struck by lightning take a considerable amount of time, even minor damage often leads to delays to the next flight. Furthermore, major damage can lead to flight cancellations, significantly affecting flight schedules.

Aircraft operations are broadly divided into the cruising phase and the takeoff and landing phase, with separate weather information support technologies used for each phase. For lightning weather support during the cruising phase, information from a lightning monitoring system called LIDEN (LIghtning DEtection Network system) operated by the Japan Meteorological Agency is widely used. In addition, because aircraft are more likely to take evasive action while cruising, lightning strikes do not occur often during the cruising phase. However, it is estimated that more than 90% of all lightning strikes occur during the takeoff and landing phase.

In relation to such lightning strikes on aircraft, technology is known for quantitatively detecting the threat of lightning (see, for example, Patent Documents 1 and 2).

JP 2010-241412 A JP 2019-45403 A

However, conventional technology has not fully considered how to reduce the number of lightning strikes on aircraft. Furthermore, the above-mentioned issues are not limited to aircraft, but are also common to other moving objects such as ships and automobiles.

The present invention has been made in consideration of the above circumstances, and an object of the present invention is to provide an information processing device, an information processing method, and a program that can reduce the number of lightning strikes on moving objects.

One aspect of the present invention is an information processing device that includes an acquisition unit that acquires threat level information indicating the threat level of lightning in the target space at the observation time or a time in the future from the observation time, the threat level being derived based on at least one of meteorological observation data representing the weather in a target space observed at a certain observation time and meteorological forecast data representing the weather in the target space predicted by a weather forecast model; a generation unit that generates multiple routes along which a moving body should move in the target space; an evaluation unit that evaluates each of the multiple routes generated by the generation unit based on the threat level information acquired by the acquisition unit; and a selection unit that selects one route from the multiple routes based on the evaluation result by the evaluation unit.

According to one aspect of the present invention, the number of lightning strikes on moving objects can be reduced.

FIG. 1 is a diagram illustrating an example of a configuration of an information processing system according to an embodiment. FIG. 2 is a diagram illustrating an example of a configuration of a first information processing device according to an embodiment. 11 is a flowchart showing an example of a flow of a series of processes in a first information processing device according to the embodiment. FIG. 13 is a diagram illustrating an example of a threat level map. FIG. 2 is a diagram illustrating an example of a configuration of a second information processing device according to an embodiment. 10 is a flowchart showing an example of a flow of a series of processes of a second information processing device according to the embodiment. FIG. 2 is a diagram showing an example of multiple flight paths. FIG. 1 is a diagram illustrating an example of a generative model. FIG. 4 is a diagram showing an example of a screen of a display unit. FIG. 1 illustrates an example of a configuration of a learning device according to an embodiment. 11 is a flowchart showing an example of a flow of a series of processes of the learning device according to the embodiment.

Hereinafter, an information processing device, an information processing method, and a program according to an embodiment of the present invention will be described with reference to the drawings.

[Configuration of Information Processing System]
1 is a diagram showing an example of a configuration of an information processing system 1 according to an embodiment. As shown in the figure, the information processing system 1 includes, for example, a weather observation device 10, a weather forecasting device 20, a first information processing device 100, a second information processing device 200, and a learning device 300. These devices are connected to a network NW. The network NW is, for example, a wide area network (WAN) or a local area network (LAN).

The weather observation device 10 observes the weather in the target space to be observed (hereinafter referred to as the observation space) using various sensors such as weather radar and radiosondes, and generates data (hereinafter referred to as the weather observation data) that represents the observation results. The observation space includes the destination to which the moving object is to move and the space nearby. The observation space may also include the travel path that the moving object passes through before reaching the destination. The moving object may be, for example, an aircraft, a ship, a rocket, an automobile, or a railroad car. For example, if the moving object is an aircraft, the observation space may include an airport and the flight path to the airport. The following description will be given assuming that the moving object is an aircraft as an example.

The weather observation device 10 may be installed, for example, on the grounds of an airport where aircraft take off and land, or in the vicinity thereof, or may be mounted on the aircraft. The weather observation data includes physical quantities that indicate atmospheric conditions, such as echo intensity (precipitation intensity), Doppler velocity (wind speed), wind direction, temperature, and humidity. When the observation space is divided into multiple grids (also called meshes), the physical quantities may be associated with each of the multiple grids. The grids may be divided into a square lattice shape with intervals of, for example, 5 km or 20 km.

The weather forecasting device 20 predicts future weather in the observation space from the weather observation data generated by the weather observation device 10 based on, for example, a weather forecast model (also called a numerical forecast model), and generates data representing the prediction results (hereinafter referred to as weather forecast data). The weather forecast data may include physical quantities of the same type as those included in the weather observation data, such as precipitation intensity, wind speed, wind direction, temperature, and humidity. The physical quantities in the weather forecast data may be associated with each of a number of grids that divide the observation space, similar to the weather observation data.

The first information processing device 100 is installed, for example, within the grounds of an airport. The first information processing device 100 acquires meteorological observation data from the meteorological observation device 10 and acquires meteorological forecast data from the weather forecast device 20 via the network NW. The first information processing device 100 then quantitatively analyzes the threat of lightning in the observation space based on either or both of the acquired meteorological observation data and weather forecast data.

The second information processing device 200 is installed, for example, inside an aircraft. The second information processing device 200 acquires the analysis results from the first information processing device 100 via the network NW. Then, based on the acquired analysis results, the second information processing device 200 provides various information to the pilot, flight attendants, etc. of the aircraft and controls the aircraft.

The learning device 300 is a device that learns (trains) the generative model MDL. Details will be described later.

[Configuration of first information processing device]
The configuration of the first information processing device 100 will be described below. The first information processing device 100 may be a single device, or may be a system in which multiple devices connected via a network NW operate in cooperation with each other. That is, the first information processing device 100 may be implemented by multiple computers (processors) included in a system using distributed computing or cloud computing.

FIG. 2 is a diagram showing an example of the configuration of the first information processing device 100 according to an embodiment. As shown in the figure, the first information processing device 100 includes, for example, a communication unit 102, a display unit 104, a control unit 110, and a storage unit 130.

The communication unit 102 includes, for example, a NIC (Network Interface Card) and a wireless communication module including a receiver and a transmitter. The communication unit 102 communicates with the weather observation device 10, the weather forecasting device 20, the second information processing device 200, the learning device 300, etc., via the network NW.

The display unit 104 is a user interface that displays various types of information. For example, the display unit 104 displays an image generated by the control unit 110. The display unit 104 may also display a GUI (Graphical User Interface) for accepting various input operations from a user (e.g., an airport staff member, etc.). For example, the display unit 104 is an LCD (Liquid Crystal Display) or an organic EL (Electroluminescence) display, etc.

The control unit 110 includes, for example, an acquisition unit 112, a derivation unit 114, a prediction unit 116, a map generation unit 118, and an output control unit 120.

The components of the control unit 110 are realized, for example, by a processor such as a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit) executing a program stored in the storage unit 130. In addition, some or all of the components of the control unit 110 may be realized by hardware such as an LSI (Large Scale Integration), an ASIC (Application Specific Integrated Circuit), or an FPGA (Field-Programmable Gate Array), or may be realized by a combination of software and hardware.

The storage unit 130 is realized by, for example, a hard disk drive (HDD), a flash memory, an electrically erasable programmable read only memory (EEPROM), a read only memory (ROM), a random access memory (RAM), etc. Various programs such as firmware and application programs are stored in the storage unit 130. In addition to the programs referenced by the processor, the storage unit 130 also stores a threat level map D1 and the like. The threat level map D1 will be described later.

[Processing flow of the first information processing device]
Hereinafter, a series of processing flows of the first information processing device 100 will be described with reference to the flowchart. FIG. 3 is a flowchart showing an example of a series of processing flows of the first information processing device 100 according to the embodiment. The processing of this flowchart may be repeated at a predetermined cycle, for example. In addition, when the first information processing device 100 is implemented by a plurality of computers included in a system using distributed computing or cloud computing, a part or all of the processing of this flowchart may be processed in parallel by the plurality of computers.

First, the acquisition unit 112 acquires meteorological observation data from the meteorological observation device 10 and acquires meteorological forecast data from the weather forecasting device 20 via the communication unit 102 (step S100). The acquisition unit 112 may acquire only one of the meteorological observation data and the meteorological forecast data.

Next, the derivation unit 114 derives an index value (hereinafter referred to as the threat level) indicating the degree of the threat of lightning in the observation space (e.g., near an airport or a flight path) at the observation time based on one or both of the meteorological observation data and the meteorological forecast data acquired by the acquisition unit 112 (step S102). The observation time may be the time when the meteorological observation device 10 observed the weather in the observation space, or it may be a future time when the weather forecasting device 20 predicted the weather in the observation space from the meteorological observation data.

The threat level may be a discrete value, such as a "low value (0.00 to 0.33)", a "medium value (0.33 to 0.66)", or a "high value (0.66 to 1.00)". The maximum value of the threat level is not limited to 1 and may be any value.

For example, the derivation unit 114 derives the probability of an aircraft landing or passing through the observation space being struck by lightning, and the time loss (e.g., flight delay time, etc.) and economic loss (e.g., damage to the aircraft, restoration costs, and flight delay time, etc.) caused by lightning strike as the threat level, using the method described in Patent Document 2. More specifically, the derivation unit 114 uses a certain function f(x) to derive the probability of lightning strike occurrence, time loss, economic loss, etc. as the threat level of lightning. The threat level of lightning may also be interpreted as the impact on human economic activity when an aircraft is struck by lightning.

The function f(x) is a function that outputs the threat level of lightning (probability of lightning strike, time loss, economic loss, etc.) as a response variable when each physical quantity included in the meteorological observation data or weather forecast data is input as an explanatory variable x, and may be a linear function such as f(x) = a1x1 + a2x2 + ..., where a1 and a2 are weighting coefficients. The function f(x) may also include a bias component. The weighting coefficient and bias component may be determined by the least squares method or the like, based on the meteorological observation data or weather forecast data observed when an aircraft is actually struck by lightning, and the time loss and economic loss caused by the lightning strike, for example. The function f(x) may also be implemented by a neural network as described in Patent Document 2.

In this way, the derivation unit 114 inputs a multidimensional vector or tensor whose elements are the physical quantities contained in the meteorological observation data and the weather forecast data as explanatory variables x to the function f(x), and derives the value output by the function f(x) as the lightning threat level. When the observation space is divided into multiple grids, the derivation unit 114 derives the lightning threat level of each grid from the physical quantities associated with each of the multiple grids.

Next, the prediction unit 116 predicts the threat level of lightning in the observation space at a future time from the threat level of lightning derived by the derivation unit 114 using an extrapolation method or the like (step S104). If the observation space is divided into multiple grids, the prediction unit 116 may predict the future threat level of lightning for each grid.

Next, the map generating unit 118 generates data in which the lightning threat level is associated with each grid based on one or both of the lightning threat level derived by the derivation unit 114 and the lightning threat level predicted by the prediction unit 116, as a threat level map D1 (step S106). The map generating unit 118 stores the generated threat level map D1 in the memory unit 130. The threat level map D1 is an example of "threat level information".

Figure 4 shows an example of a threat level map D1. As shown in the figure, the threat level map D1 may be image data in which the lightning threat level of each grid is replaced with pixel values such as brightness, saturation, and hue. Such a threat level map D1 may exist for each observation time.

Returning to the explanation of the flowchart in FIG. 3, the output control unit 120 then transmits the threat level map D1 generated by the map generation unit 118 to the second information processing device 200 via the communication unit 102, and displays it on the display unit 104 (step S108). This completes the processing of this flowchart.

[Configuration of second information processing device]
The following describes the configuration of the second information processing device 200. Fig. 5 is a diagram showing an example of the configuration of the second information processing device 200 according to an embodiment. As shown in the figure, the second information processing device 200 includes, for example, a communication unit 202, a display unit 204, a control unit 210, and a storage unit 230.

The communication unit 202 includes, for example, a NIC and a wireless communication module including a receiver and a transmitter. The communication unit 202 communicates with the first information processing device 100, the learning device 300, etc., via the network NW.

The display unit 204 is a user interface that displays various types of information. For example, the display unit 204 displays an image generated by the control unit 210. The display unit 204 may also display a GUI for accepting various input operations from, for example, an aircraft pilot. For example, the display unit 204 is an LCD or an organic EL display. The display unit 204 is an example of an "output unit."

The control unit 210 includes, for example, an acquisition unit 212, a route generation unit 214, a route evaluation unit 216, a route adjustment unit 218, a route selection unit 220, an output control unit 222, and a flight control unit 224. The flight control unit 224 is an example of a "mobile control unit."

The components of the control unit 210 are realized, for example, by a processor such as a CPU or GPU executing a program stored in the storage unit 230. In addition, some or all of the components of the control unit 210 may be realized by hardware such as an LSI, ASIC, or FPGA, or may be realized by a combination of software and hardware.

The storage unit 230 is realized by, for example, a HDD, a flash memory, an EEPROM, a ROM, a RAM, etc. The storage unit 230 stores various programs such as firmware and application programs. In addition to the programs referenced by the processor, the storage unit 230 also stores the above-mentioned threat level map D1 and the generative model data D2. The generative model data D2 is information (program or data structure) that defines the generative model MDL described later. This generative model data D2 may be installed in the storage unit 230 from the learning device 300 via the network NW, for example. In addition, when a portable storage medium storing the generative model data D2 is connected to the drive device of the second information processing device 200, the generative model data D2 may be installed in the storage unit 230 from the portable storage medium.

[Processing flow of the second information processing device]
Hereinafter, a series of processing steps of the second information processing device 200 will be described with reference to the flowchart. Fig. 6 is a flowchart showing an example of a series of processing steps of the second information processing device 200 according to the embodiment. The processing steps of this flowchart may be repeated at a predetermined interval, for example.

First, the acquisition unit 212 acquires the threat level map D1 from the first information processing device 100 via the communication unit 202 (step S200).

Next, the route generation unit 214 generates multiple flight routes for the aircraft (i.e., the aircraft itself) on which the second information processing device 200 is mounted (step S202).

Figure 7 is a diagram showing an example of multiple flight routes. In the figure, Gx represents a grid in which an airport, which is the destination of the aircraft, is located. For example, the route generation unit 214 generates one flight route R1 that connects the current position of the aircraft to the destination airport as a predetermined standard flight route. Next, the route generation unit 214 randomly generates multiple flight routes such as flight routes R2 and R3 from the flight route R1 using random numbers or the like. Note that the route generation unit 214 may generate only a single flight route R1 without generating flight routes R2 and R3.

The route generation unit 214 may also generate the flight path of the aircraft using the generative model MDL defined by the generative model data D2.

Figure 8 is a diagram showing an example of the generation model MDL. As shown in the example, the generation model MDL is a model that has been trained to output a flight path that an aircraft should travel in the observation space at a certain time tx when a threat level map D1 that represents the threat level of lightning in the observation space at the same time tx is input.

For example, the route generation unit 214 inputs the threat level map D1 acquired by the acquisition unit 212 into the learned (trained) generation model MDL, and sets the flight route output by the generation model MDL as the travel route that the aircraft should travel at the current observation time.

The generation model MDL may be trained to output the flight path of an aircraft when other types of data are input in addition to the threat level map D1. The other types of data may be, for example, data indicating the presence or absence of lightning and the number of lightning occurrences. The other types of data may be, for example, data indicating the degree of damage to an aircraft that has been hit by lightning, such as the degree of damage to the aircraft and the cost of restoration.

Such a generative model MDL may be implemented by various models, such as a neural network, a support vector machine, regularized regression, a random forest, or a Gaussian process regression. In the following, as an example, the generative model MDL will be described as being implemented by a neural network.

When the generative model MDL is implemented by a neural network, the generative model data D2 includes various information, such as, for example, coupling information on how the units included in each of the multiple layers that make up the neural network are coupled to each other, and coupling coefficients assigned to data input and output between coupled units.

The coupling information includes, for example, the number of units included in each layer, information specifying the type of unit to which each unit is coupled, the activation function of each unit, and information on gates provided between units in the hidden layer. The activation function may be, for example, a normalized linear function (ReLU function), a sigmoid function, a step function, or other functions. The gate selectively passes or weights data transmitted between units, for example, depending on the value (e.g., 1 or 0) returned by the activation function. The coupling coefficient includes, for example, a weight assigned to output data when data is output from a unit in a layer to a unit in a deeper layer in the hidden layer of a neural network. The coupling coefficient may also include a bias component specific to each layer.

Returning to the explanation of the flowchart in FIG. 6, the route evaluation unit 216 then evaluates each of the multiple flight routes generated by the route generation unit 214 based on the threat level map D1 acquired by the acquisition unit 212 (step S204).

For example, the route evaluation unit 216 evaluates each flight route based on a certain evaluation function. The evaluation function is, for example, a function with the threat level of lightning and the operational efficiency of the aircraft as explanatory variables. The operational efficiency of the aircraft includes, for example, the travel time required when the aircraft flies along the flight route generated by the route generation unit 214, and the amount of energy (fuel consumption) consumed when the aircraft flies along the flight route generated by the route generation unit 214. In other words, the evaluation function may be a function with the threat level of lightning, the travel time, the amount of energy consumed, etc. as explanatory variables.

For example, the evaluation function outputs a smaller evaluation value the greater the threat of lightning in the grids along the flight path, and outputs a smaller evaluation value the lower the operational efficiency of the aircraft (the longer the flight time or the higher the energy consumption). Conversely, the evaluation function outputs a larger evaluation value the smaller the threat of lightning in the grids along the flight path, and outputs a larger evaluation value the higher the operational efficiency of the aircraft (the shorter the flight time or the lower the energy consumption).

For example, among the multiple flight routes generated by the route generation unit 214, the route evaluation unit 216 evaluates a flight route with a larger evaluation value returned by the evaluation function more highly, and evaluates a flight route with a smaller evaluation value returned by the evaluation function less highly.

The route evaluation unit 216 may also evaluate the flight route in more detail from the standpoint of safety and cost. For example, the route evaluation unit 216 may evaluate a flight route that has a high evaluation value due to a low threat of lightning as a safe flight route. The route evaluation unit 216 may also evaluate a flight route that has a high evaluation value due to a short flight time or low energy consumption as a low-cost flight route.

When the route generation unit 214 generates the flight route of the aircraft using the generative model MDL, the process of S204 may be omitted. For example, flight history data of aircraft that were not struck by lightning or were struck by lightning but suffered small time losses or economic losses may be extracted from the past flight history data of an unspecified number of aircraft, and the generative model MDL may be trained in advance using the extracted flight history data. In such a case, it is expected that the generative model MDL has already learned the evaluation function. In other words, the generative model MDL may generate only flight routes with high evaluations. Therefore, when the flight route of the aircraft is generated using the generative model MDL, the route evaluation unit 216 may omit evaluation of the flight route.

Next, the route adjustment unit 218 adjusts the flight route with the highest evaluation among the multiple flight routes evaluated by the route evaluation unit 216 so as to further improve the evaluation value of the evaluation function (step S206). In addition, the route adjustment unit 218 may adjust not only the flight route with the highest evaluation, but also other flight routes evaluated by the route evaluation unit 216 so as to improve the evaluation value of the evaluation function.

For example, the route adjustment unit 218 may use an optimization algorithm to adjust the latitude, longitude, altitude, etc. of the flight route. The optimization algorithm may be, for example, a gradient method or a genetic algorithm. The gradient method may be, for example, a steepest descent method or a stochastic gradient descent method.

Next, the route selection unit 220 selects an optimal flight route from among the multiple flight routes (including the flight route adjusted by the route adjustment unit 218) based on the evaluation results of the multiple flight routes evaluated by the route evaluation unit 216 (step S208). For example, the route selection unit 220 may select the flight route with the largest evaluation value of the evaluation function among the multiple flight routes as the optimal flight route.

Next, the output control unit 222 causes information representing the optimal flight route selected by the route selection unit 220 to be displayed as an image on the display unit 204 or output as sound from a speaker (not shown) (step S210).

The output control unit 222 may also transmit information indicating the optimal flight route to the first information processing device 100 at the airport via the communication unit 202. When the information indicating the optimal flight route is received by the communication unit 102, the output control unit 120 of the first information processing device 100 may display the information on the display unit 104. This makes it possible to inform airport air traffic controllers and the like of the optimal flight route.

The output control unit 222 may also cause information representing each of the multiple flight routes to be selected by the route selection unit 220 to be displayed as an image on the display unit 204 or output as sound from a speaker. At this time, the output control unit 222 may also output the evaluation results of the route evaluation unit 216 for each flight route. The speaker is another example of an "output unit".

FIG. 9 is a diagram showing an example of the screen of the display unit 204. As shown in the example, the output control unit 222 may display an image on the display unit 204 in which the evaluation results are associated with each of the flight routes R1, R2, and R3. In the example shown, the evaluation result of the flight route R1, which has the longest flight distance to reach the destination (grid Gx), represents a medium rank in safety and a high rank in cost. The evaluation result of the flight route R2, which has a shorter flight distance than the flight route R1, represents a high rank in safety and a medium rank in cost. The evaluation result of the flight route R3, which has the shortest flight distance, represents a medium rank in safety and a low rank in cost. In this way, multiple flight routes may be ranked from the perspective of safety and cost.

For example, when a screen such as that shown in FIG. 9 is displayed on the display unit 204, an aircraft pilot may refer to the evaluation results and select one flight route from among multiple flight routes based on experience and visually confirmed weather conditions in the sky. That is, an operation to select a flight route may be input to the input unit 206. In this case, the route selection unit 220 may select one flight route from among multiple flight routes by preferentially referring to the input operation to the input unit 206 over the evaluation results of the route evaluation unit 216.

For example, suppose that flight route R2 has the highest evaluation value among flight routes R1, R2, and R3, and the pilot selects flight route R3 instead of flight route R2. In this case, instead of selecting flight route R2 with the highest evaluation value as the optimal flight route, the route selection unit 220 selects flight route R3 selected by the pilot as the optimal flight route. In this way, the route selection unit 220 may interactively select the optimal flight route. This type of operation is particularly useful when it is determined that a route should be selected based on criteria other than a predetermined evaluation function based on the surrounding circumstances of the scene. The surrounding circumstances of the scene are, for example, a situation in which another major risk that has not been anticipated occurs.

Returning to the explanation of the flowchart in FIG. 6, the flight control unit 224 then flies the aircraft autonomously or on autopilot, based on the optimal flight route selected by the route selection unit 220 (step S212). Autonomous is a mode in which the aircraft flies automatically without intervention by the pilot. Autopilot is a mode in which the aircraft flies automatically with some intervention by the pilot. This completes the processing of this flowchart.

[Configuration of learning device]
The configuration of the learning device 300 will be described below. The learning device 300 may be a single device, similar to the first information processing device 100, or may be a system in which multiple devices connected via a network NW operate in cooperation with each other. In other words, the learning device 300 may be implemented by multiple computers (processors) included in a system that uses distributed computing or cloud computing.

FIG. 10 is a diagram showing an example of the configuration of a learning device 300 according to an embodiment. As shown in the figure, the learning device 300 includes, for example, a communication unit 302, a control unit 310, and a storage unit 330.

The communication unit 302 includes, for example, a NIC and a wireless communication module including a receiver and a transmitter. The communication unit 302 communicates with the first information processing device 100, the second information processing device 200, etc. via the network NW.

The control unit 310 includes, for example, an acquisition unit 312 and a learning unit 314. The components of the control unit 310 are realized, for example, by a processor such as a CPU or GPU executing a program stored in the storage unit 330. In addition, some or all of the components of the control unit 310 may be realized by hardware such as an LSI, ASIC, or FPGA, or may be realized by a combination of software and hardware.

The storage unit 330 is realized by, for example, a HDD, a flash memory, an EEPROM, a ROM, a RAM, etc. The storage unit 330 stores various programs such as firmware and application programs. In addition to the programs referenced by the processor, the storage unit 330 also stores the above-mentioned generative model data D2 and teacher data D3, etc.

The teacher data D3 is data for learning (training) the generation model MDL. For example, the teacher data D3 is a data set in which a threat level map D1 representing the threat level of lightning in an observation space at a certain time tx is associated with a flight path in which an aircraft was not struck by lightning in the observation space at the same time tx as a teacher label (also called a target). The teacher data D3 may also be a data set in which a flight path in which an aircraft was not struck by lightning is associated with a combination of the threat level map D1 and other types of data as a teacher label. As described above, the other types of data are data representing the presence or absence of lightning, the number of occurrences, the degree of damage to aircraft that have been struck by lightning, etc. Such teacher data D3 may be extracted from past flight history data of an unspecified number of aircraft.

[Learning device processing flow]
The flow of a series of processes of the learning device 300 will be described below with reference to the flowchart. FIG. 11 is a flowchart showing an example of a series of processes of the learning device 300 according to the embodiment. The process of this flowchart may be repeated at a predetermined cycle, for example. In addition, when the learning device 300 is implemented by multiple computers included in a system using distributed computing or cloud computing, some or all of the process of this flowchart may be processed in parallel by the multiple computers.

First, the acquisition unit 312 acquires the threat level map D1, which is input data, from the teacher data D3 stored in the memory unit 330 (step S300).

Next, the learning unit 314 inputs the threat level map D1 acquired by the acquisition unit 312 into the unlearned generative model MDL (step S302).

Next, the learning unit 314 obtains the flight path output by the generative model MDL from the generative model MDL (step S304).

Next, the learning unit 314 calculates the difference (also called the loss) between the flight path obtained from the generation model MDL and the flight path that was associated as a teacher label in the threat level map D1 input to the generation model MDL (step S306).

Next, the learning unit 314 learns the generative model MDL so as to reduce the calculated difference (step S308). For example, the learning unit 314 may determine (update) parameters of the generative model MDL, such as weight coefficients and bias components, using a stochastic gradient descent method or the like so as to reduce the difference.

The learning unit 314 stores the learned generative model MDL in the memory unit 230 as generative model data D2.

In this way, the learning unit 314 repeats (iterates) the processes from S300 to S308 to learn the generative model MDL. Then, the learning unit 314 transmits the generative model data D2 that defines the sufficiently learned generative model MDL to the second information processing device 200, for example, via the communication unit 302. This ends the processing of this flowchart.

According to the embodiment described above, the second information processing device 200 acquires a threat level map D1 generated based on at least one of meteorological observation data and weather forecast data of the observation space at a certain observation time. The second information processing device 200 generates a plurality of flight paths along which the aircraft should fly in the observation space. The second information processing device 200 evaluates each of the generated flight paths based on the acquired threat level map D1. The second information processing device 200 selects an optimal flight path from among the multiple flight paths based on the evaluation results of each of the multiple flight paths. Furthermore, the second information processing device 200 provides information representing the selected optimal flight path to the pilot of the aircraft or an airport controller, or controls the aircraft based on the optimal flight path. This can reduce the number of lightning strikes on aircraft. As a result, the time loss and economic loss of the aircraft due to lightning strikes can be reduced.

The above describes the form for carrying out the present invention using an embodiment, but the present invention is not limited to such an embodiment, and various modifications and substitutions can be made without departing from the spirit of the present invention.

1...information processing system, 10...weather observation device, 20...weather forecasting device, 100...first information processing device, 102...communication unit, 104...display unit, 110...control unit, 112...acquisition unit, 114...derivation unit, 116...prediction unit, 118...map generation unit, 120...output control unit, 130...storage unit, 200...second information processing device, 202...communication unit, 204...display unit, 206...input unit, 210...control unit, 212...acquisition unit, 214...route generation unit, 216...route evaluation unit, 218...route adjustment unit, 220...route selection unit, 222...output control unit, 224...flight control unit, 300...learning device, 302...communication unit, 310...control unit, 312...acquisition unit, 314...learning unit, 330...storage unit

Claims (11)

an acquisition unit that acquires threat level information that indicates the threat level of lightning in the target space at the observation time or at a time in the future from the observation time, the threat level being derived based on at least one of meteorological observation data that indicates the weather in the target space observed at a certain observation time and meteorological forecast data that indicates the weather in the target space at a time in the future from the observation time predicted from the meteorological observation data at the observation time by a weather forecast model;
A generation unit that generates a plurality of paths along which a moving object should move in the target space;
an evaluation unit that evaluates each of the plurality of paths generated by the generation unit based on the threat level information acquired by the acquisition unit;
a selection unit that selects one path from the plurality of paths based on an evaluation result by the evaluation unit,
the generation unit inputs the threat level information acquired by the acquisition unit into a trained model, which is a machine learning model trained based on teacher data in which a route that the moving body should move in the target space at a certain target time is associated with threat level information indicating a threat level of lightning in the target space at the target time, and generates a plurality of routes that the moving body should move in the target space at the observation time based on an output result of the trained model to which the threat level information is input;
the target space is partitioned by a plurality of grids;
the threat level information is a threat level map which is image data in which the threat level of lightning associated with each of the plurality of grids is replaced with a predetermined pixel value;
The teacher data is a data set in which a route in which the moving object is not struck by lightning in the target space at the target time is associated with the threat level map at least at the target time.
Information processing device. the evaluation unit evaluates each of the plurality of routes based on an evaluation function having the threat level and the operational efficiency of the mobile object as explanatory variables.
The information processing device according to claim 1 . The evaluation function is a function that evaluates the route lower as the threat level is higher or the operational efficiency of the mobile unit is lower, and evaluates the route higher as the threat level is lower or the operational efficiency of the mobile unit is higher.
The information processing device according to claim 2 . The operational efficiency of the moving body includes a travel time required for the moving body to move along the route and an amount of energy consumed when the moving body moves along the route.
4. The information processing device according to claim 2 or 3 . A movement control unit that controls movement of the moving object based on the route selected by the selection unit.
The information processing device according to claim 1 . an output unit that outputs information;
and an output control unit that causes the output unit to output information representing the route selected by the selection unit.
The information processing device according to claim 1 . further comprising an input unit for inputting an input operation by a user;
The output control unit causes the output unit to output information representing each of the plurality of paths evaluated by the evaluation unit;
when an operation for selecting one of the plurality of routes is input to the input unit after the output unit outputs information representing each of the plurality of routes, the selection unit preferentially refers to the input operation to the input unit rather than the evaluation result of the evaluation unit and selects one route from the plurality of routes.
The information processing device according to claim 6 . The output control unit causes the output unit to output information in which the evaluation result of the evaluation unit is associated with each path.
The information processing device according to claim 7 . an acquisition unit that acquires threat level information that indicates the threat level of lightning in the target space at the observation time or at a time in the future from the observation time, the threat level being derived based on at least one of meteorological observation data that indicates the weather in the target space observed at a certain observation time and meteorological forecast data that indicates the weather in the target space at a time in the future from the observation time predicted from the meteorological observation data at the observation time by a weather forecast model;
a generation unit that inputs the threat level information acquired by the acquisition unit into a trained model, which is a machine learning model trained based on teacher data in which a route that a moving body should move in the target space at a certain target time is associated with threat level information indicating a threat level of lightning in the target space at the certain target time, and generates a route that the moving body should move in the target space at the observation time based on an output result of the trained model to which the threat level information is input ,
the target space is partitioned by a plurality of grids;
the threat level information is a threat level map which is image data in which the threat level of lightning associated with each of the plurality of grids is replaced with a predetermined pixel value;
The teacher data is a data set in which a route in which the moving object is not struck by lightning in the target space at the target time is associated with the threat level map at least at the target time.
Information processing device. The computer
obtaining threat level information indicating a threat level of lightning in the target space at the observation time or at a time in the future from the observation time, the threat level being derived based on at least one of meteorological observation data indicating the weather in the target space observed at a certain observation time and meteorological forecast data indicating the weather in the target space at a time in the future from the observation time predicted from the meteorological observation data at the observation time by a weather forecast model;
generating a plurality of routes along which a moving object should move in the target space;
Evaluating each of the generated routes based on the acquired threat level information;
selecting one route from among the plurality of routes based on the evaluation results of each of the plurality of routes;
inputting the acquired threat level information into a trained model, which is a machine learning model trained based on teacher data in which a route that the moving body should move in the target space at a certain target time is associated with threat level information indicating a threat level of lightning in the target space at the certain target time;
generating a plurality of routes along which the moving object should move in the target space at the observation time based on an output result of the trained model to which the threat level information has been input;
the target space is partitioned by a plurality of grids;
the threat level information is a threat level map which is image data in which the threat level of lightning associated with each of the plurality of grids is replaced with a predetermined pixel value;
The teacher data is a data set in which a route in which the moving object is not struck by lightning in the target space at the target time is associated with the threat level map at least at the target time.
Information processing methods. A program for causing a computer to execute the program,
acquiring threat level information which is derived based on at least one of meteorological observation data representing the weather in a target space observed at a certain observation time and meteorological forecast data representing the weather in the target space at a time in the future from the observation time predicted from the meteorological observation data at the observation time by a weather forecast model, and which indicates the threat level of lightning in the target space at the observation time or at a time in the future from the observation time;
generating a plurality of routes along which a moving object should move in the target space;
Evaluating each of the generated multiple routes based on the acquired threat level information;
selecting one route from the plurality of routes based on an evaluation result of each of the plurality of routes;
inputting the acquired threat level information into a trained model, which is a machine learning model trained based on teacher data in which a route that the moving body should move in the target space at a certain target time is associated with threat level information indicating a threat level of lightning in the target space at the target time;
generating a plurality of routes along which the moving object should move in the target space at the observation time based on an output result of the trained model to which the threat level information has been input ;
the target space is partitioned by a plurality of grids;
the threat level information is a threat level map which is image data in which the threat level of lightning associated with each of the plurality of grids is replaced with a predetermined pixel value;
The teacher data is a data set in which a route in which the moving object is not struck by lightning in the target space at the target time is associated with the threat level map at least at the target time.
program .

Images