JP7457217B2 - information processing equipment - Google Patents

Description

The present invention relates to an information processing device capable of communicating with a plurality of navigation terminals and capable of predicting specific matters.

When a navigation terminal recognizes the current position using a GPS signal or the like and inputs a destination, the optimal route, required time, estimated arrival time, etc. are displayed (for example, Patent Document 1). Most navigation terminals are installed in vehicles, but mobile terminals (smartphones, etc.) with navigation apps are sometimes used as navigation terminals.

Furthermore, navigation terminals in recent years have communication functions and can reflect traffic information (congestion information), etc. This improves the accuracy of the predicted arrival time.

Furthermore, the current location is recognized even while moving, and the predicted arrival time is updated at any time. This further improves the accuracy of the predicted arrival time.

Incidentally, in recent years, big data has been utilized in various fields to predict demand based on learning models (for example, Patent Document 2).

Patent No. 4725637 Japanese Patent Application Publication No. 2019-128922

For example, a learning model is constructed based on the correlation between input parameter data, such as date and time information (past day of the week, time, whether it was a holiday or not), and weather information (temperature, precipitation, etc.), and actual measured data for a specific item. By inputting the prediction parameters for the time to be predicted into the learning model, a prediction for the specific item is output. This type of prediction system is highly versatile, as it can output rough prediction results simply by inputting the prediction parameters, without verifying causal relationships.

However, predictions based on past data have limited prediction accuracy.

On the other hand, the predicted arrival time in a navigation system including a navigation terminal is highly accurate and very useful. Furthermore, the prediction is based on relatively recent circumstances, which is also useful. However, the predicted arrival time is only presented to the user of the navigation terminal, and there are not many examples of it being effectively utilized.

Therefore, the inventor of the present invention considered improving the prediction accuracy by utilizing the predicted arrival time in predicting specific items.

The present invention solves the above problems, and aims to provide a prediction system with high prediction accuracy.

An information processing device of the present invention that solves the above problems includes a communication unit capable of communicating with a plurality of navigation terminals, and a communication unit that transmits information such as a destination input to the navigation terminal, a position at the time of input, and a communication unit capable of communicating with a plurality of navigation terminals. a terminal input information acquisition unit that acquires terminal input information including the time when the navigation terminal arrives; an arrival time prediction unit that predicts the time at which the navigation terminal will arrive at a specific position including the destination; and a prediction generated by machine learning. a specific matter prediction unit that applies a model to predict specific matters at the specific position; and a specific matter that is performed by the specific matter prediction unit based on the arrival time of the navigation terminal predicted by the arrival time prediction unit. and a modification unit that modifies the prediction information related to the prediction information.

As a result, it is possible to roughly predict specific matters by the day before, and the accuracy of the prediction is improved by making corrections on the same day.

An information processing device of the present invention that solves the above problems includes a communication unit capable of communicating with a plurality of navigation terminals, and a communication unit that transmits information such as a destination input to the navigation terminal, a position at the time of input, and a communication unit capable of communicating with a plurality of navigation terminals. a terminal input information acquisition unit that acquires terminal input information including the time at which the navigation terminal is located; and a specific position where the navigation terminal is provided with a display device that can switch and display guidance information to the destination and information other than the guidance information. A passing time prediction unit that predicts the time when the vehicle will pass through the destination, and a traffic volume estimation model generated by machine learning are applied to display the guidance information in accordance with the traffic volume time distribution from the specific location to the destination. displaying the guidance information set by the guidance information display time slot setting unit based on the passage time of the navigation terminal predicted by the passage time prediction unit; A correction unit for correcting the time zone.

Accurate predictions allow drivers to be guided reliably and can increase advertising revenue.

An information processing device of the present invention that solves the above problems includes a communication unit capable of communicating with a plurality of navigation terminals, and a communication unit that transmits information such as a destination input to the navigation terminal, a position at the time of input, and a communication unit capable of communicating with a plurality of navigation terminals. a terminal input information acquisition unit that acquires terminal input information including the time when the navigation terminal arrives at the destination; an arrival time prediction unit that predicts the time when the navigation terminal will arrive at the destination; and a prediction model generated by machine learning. a cooking schedule creation unit that creates a cooking schedule for food to be served at the destination; and a cooking schedule created by the cooking schedule creation unit based on the arrival time of the navigation terminal predicted by the arrival time prediction unit. and a correction section for making corrections.

Accurate predictions allow customers to serve freshly prepared food with almost no waiting time, and reduce excessive food waste.

Preferably, in the above invention, the terminal input information includes a terminal ID of the navigation terminal, and further includes a preference information acquisition unit that acquires preference information stored for each terminal ID, and the modification unit includes: The cooking schedule created by the cooking schedule creation section is corrected based on the arrival time of the navigation terminal predicted by the arrival time prediction section and the preference information.

By reflecting preference information, a cooking schedule can be generated taking into account the ordered dishes.

An information processing device of the present invention that solves the above problems includes a communication unit capable of communicating with a plurality of navigation terminals, and a communication unit that transmits information such as a destination input to the navigation terminal, a position at the time of input, and a communication unit capable of communicating with a plurality of navigation terminals. a terminal input information acquisition unit that acquires terminal input information including the time when the navigation terminal arrives at the destination; an arrival time prediction unit that predicts the time when the navigation terminal will arrive at the destination; and a prediction model generated by machine learning. A parking situation prediction unit that predicts the parking situation of a parking lot around the destination, and parking situation prediction information created by the parking situation prediction unit based on the arrival time of the navigation terminal predicted by the arrival time prediction unit. and a correction section for making corrections.

Accurate predictions guide drivers to available parking spaces.

An information processing device of the present invention that solves the above problems includes a communication unit capable of communicating with a plurality of navigation terminals, and a communication unit that transmits a destination input to the navigation terminal, a position at the time of input, and a communication unit capable of communicating with a plurality of navigation terminals. a terminal input information acquisition unit that acquires terminal input information including the time when the navigation terminal arrives at the destination; an arrival time prediction unit that predicts the time when the navigation terminal will arrive at the destination; and a prediction model generated by machine learning. The guide placement planning unit plans the placement of guide staff in parking lots around the destination, and the guide placement planning unit plans the arrival time of the navigation terminal predicted by the arrival time prediction unit. A correction unit that corrects the guide staff placement plan information.

Accurate predictions allow for the appropriate deployment of limited guide personnel.

An information processing device of the present invention that solves the above problems includes a communication unit capable of communicating with a navigation terminal, and a destination inputted to the navigation terminal, a position at the time of input, and a time when the destination is inputted from the navigation terminal. a terminal input information acquisition unit that acquires terminal input information including the time of the navigation terminal; a passing time prediction unit that predicts the time when the navigation terminal passes a specific position; and a time period that includes the passing time predicted by the passage time prediction unit. and a guidance information transmitting section that transmits guidance information from the specific location to the destination to a navigation terminal passing through the specific location.

Accurate predictions allow drivers to be guided reliably.

In the above invention, preferably, the navigation terminal is capable of wide-area wireless communication and short-range wireless communication in a wireless receivable area, the specific position is an area where wide-area wireless communication is not possible for the navigation terminal, and the navigation terminal The information transmitter transmits guidance information via a short-range wireless communication device provided near the specific location.

This makes it possible to reliably guide the driver even if wide-area wireless communication is not possible.

An information processing device of the present invention that solves the above problems includes a communication unit capable of communicating with a navigation terminal, and a destination inputted to the navigation terminal, a position at the time of input, and a time when the destination is inputted from the navigation terminal. a terminal input information acquisition unit that acquires terminal input information including a time and a terminal ID of the navigation terminal; a passing time prediction unit that predicts a time when the navigation terminal passes a specific position; an identification unit that identifies the navigation terminal by confirming the terminal ID of the navigation terminal that passes through a specific location during a time period including the predicted passage time; A switching command transmitter is provided nearby and configured to transmit a switching command to a display device that can switch between displaying guidance information from the specific position to the destination and information other than the guidance information.

Accurate predictions allow drivers to be guided reliably and can increase advertising revenue.

The information processing device of the present invention is part of a prediction system. By using the information processing device of the present invention, improvement in prediction accuracy can be expected.

This is the schematic configuration of the prediction system. This is an example of an operation flow of an information processing device. It is an image diagram of an example in which the prediction system of the present application is applied to information signboard display switching. It is an image diagram of the basic operation of guidance signboard display switching. It is an image diagram of each passing amount data. It is an image diagram of a modification concerning guidance signboard display switching. It is an image diagram related to a guidance notification. It is an image diagram of an example in which the prediction system of the present application is applied to the cooking schedule of a large restaurant. It is an image diagram of a cooking schedule. It is an image diagram of a modification concerning cooking schedule creation. FIG. 2 is an image diagram of an example in which the prediction system of the present application is applied to predicting parking conditions in parking lots of leisure facilities such as amusement parks. It is an image diagram of parking situation prediction. FIG. 2 is an image diagram of an example in which the prediction system of the present application is applied to a guide placement plan for a parking lot of a leisure facility such as an amusement park. It is an image diagram of the guide staff allocation plan. It is an image diagram of the main points of the present application.

~Common configuration~
FIG. 1 shows a schematic configuration of a prediction system. The prediction system includes an information processing device 1 and a plurality of navigation terminals 2.

The information processing device 1, like a typical information processing device, has a processor, memory, an input/output interface, a communication interface, and a database. The database may be external. The information processing device 1 inputs information, executes a predetermined calculation using a program, and outputs the calculation results.

The information processing device 1 and the plurality of navigation terminals 2 can communicate with each other via a network.

The navigation terminal 2 is generally an in-vehicle terminal, but may also be a mobile terminal (such as a smartphone) with a built-in navigation app. The navigation terminal 2 recognizes the current location using GPS signals, etc. When a destination is input, it outputs the optimal route, required time, predicted arrival time, etc. Calculations are continued at all times while traveling, and the output results are updated. The navigation terminal 2 has a configuration equivalent to that of a typical navigation terminal.

The following describes a functional block diagram of the information processing device 1. The information processing device 1 includes a communication unit 11, a terminal information acquisition unit 12, an arrival time prediction unit 13, a specific item prediction unit 14, a prediction output unit 15, a prediction model update unit 16, and a correction unit 17.

The terminal information acquisition unit 12 and the arrival time prediction unit 13 have a similar configuration to that of a general navigation system.

The terminal information acquisition unit 12 acquires terminal information including the terminal's current location information, input destination information, and terminal identification information from the navigation terminal 2 via the communication unit 11.

The arrival time prediction unit 13 calculates the required time from the distance between the current position and the destination and the standard speed, and predicts the arrival time to the destination from the current time and the required time. Traffic information (traffic information), etc. is input from outside and corrected as appropriate.

In the above embodiment, the information processing device 1 calculates the predicted arrival time, but the terminal side or another information processing device may calculate the predicted arrival time and acquire it as one of the terminal information. That is, there is no limitation as long as the predicted arrival time can be obtained.

Furthermore, output information such as the optimal route, required time, and predicted arrival time is transmitted to the navigation terminal 2 via the communication unit 11 and displayed on the screen of the navigation terminal 2.

The specific item prediction unit 14, the prediction output unit 15, and the prediction model update unit 16 are similar in configuration to a general prediction system.

The specific matter prediction unit 14 applies a prediction model generated by machine learning to predict a specific matter that will occur at a specific location (for example, a destination). A prediction model is constructed based on the correlation between data related to a plurality of past parameters and actual measurement data of specific items. Input the corresponding parameters of the target time into the prediction model and output the prediction results related to the specific item.

The prediction output unit 15 outputs prediction results related to specific matters to external devices 3 to 5. For example, it is transmitted via the communication unit 11 to the navigation terminal 2, the mobile terminal 5 located at the location where the specific matter occurs, or another information processing device 4 installed at the location where the specific matter occurs.

The prediction model updating unit 16 receives actual measurement results regarding specific matters via the communication unit 11, and readjusts the prediction model so that the prediction approaches the actual measurement.

The modification unit 17 is a characteristic configuration of the present application. Based on the predicted arrival time by the arrival time prediction unit 13, the prediction information related to the specific matter performed by the specific item prediction unit 14 is corrected.

For example, assuming that the distribution of predicted arrival times is similar to the distribution of specific item predictions, calibration is performed so that the general shape is approximated, and the distribution of specific item predictions is corrected.

Alternatively, the predicted arrival time distribution may be input as one of the parameters of the prediction model, and the prediction result regarding the specific item may be output. In this case, the prediction model update unit 16 reuses the predicted arrival time distribution in the arrival time prediction unit 13 as learning data, analyzes the correlation with the actual measurement results, and readjusts the prediction model.

~ Common operation ~
FIG. 2 is an example of an operation flow of the information processing device 1.

The navigation terminal 2 has unique identification information. The navigation terminal 2 recognizes the current position. Further, a destination is input to the navigation terminal 2.

The terminal information acquisition unit 12 acquires terminal information including current position information of the terminal, input destination information, and terminal identification information from the navigation terminal 2 via the communication unit 11 (step 1). Note that the current time is acquired on the information processing device 1 side, the navigation terminal 2 side, or both.

The arrival time prediction unit 13 receives the terminal information from the terminal information acquisition unit 12, performs a predetermined calculation, and outputs the predicted arrival time, etc. (step 2). If necessary, the calculation result is transmitted to the navigation terminal 2 via the communication unit 11.

Meanwhile, in parallel with the operation of the navigation terminal 2, the specific event prediction unit 14 applies a prediction model generated by machine learning to make a prediction regarding a specific event occurring at a specific location (e.g., a destination) (step 3). For example, a rough prediction can be made the day before the specific event occurs. The prediction output unit 15 outputs the prediction the day before to a mobile terminal 5 or another information processing device 4 at the specific location (step 4).

On the other hand, the predicted arrival time in step 2 can be obtained on the day (for example, several hours before) that the specific event occurs.

Thereby, the modification unit 17 can reflect the most recent and accurate data (predicted arrival time) in the prediction regarding the specific item (step 5). Note that since the predicted arrival time is updated at any time, predictions regarding specific matters are also revised at any time. The prediction output unit 15 outputs the latest prediction to the mobile terminal 5 or another information processing device 4 located at a specific location.

The prediction model update unit 16 inputs the actual measurement results related to the specific items via the communication unit 11 (step 6) and readjusts the prediction model so that the predictions approach the actual measurements (step 7).

～Common effect～
Conventional prediction systems make predictions based on past actual measurement data. Although this data is associated with highly correlated parameters through machine learning, the direct causal relationship is unclear. As a result, prediction accuracy is limited.

On the other hand, when a destination is input into the navigation terminal 2, and particularly when guidance by the navigation system is continuing, there is an extremely high probability that the user (driver) of the navigation terminal 2 will reach the destination. Furthermore, the predicted arrival time is also extremely accurate. Whereas the data used in conventional prediction systems is obtained, for example, the day before or before a specific event occurs, the predicted arrival time is obtained very recently (for example, a few hours ago). In other words, there is an extremely high probability that the driver will arrive at the destination as predicted.

By correcting predictions related to specific items based on the predicted arrival time, prediction accuracy can be expected to improve.

~ Information signboard display switching ~
<Summary>
FIG. 3 is an image diagram of an example in which the prediction system of the present application is applied to information signboard display switching.

Leisure facilities in the suburbs, such as golf courses, often have information signboards at important points along the way (such as junction points). By being guided by information displays, drivers can arrive at their destination without getting lost. In recent years, most drivers use navigation systems, but sometimes the display of junctions, etc., is difficult to understand. On the other hand, when used in conjunction with this system, the driver can visually check information signboards to supplement the navigation system.

Commercial advertising signs are often placed around information signs. Criticism has also been raised that having too many signboards increases blind spots, creating safety issues, and spoiling the landscape. Therefore, it is preferable to reduce the number of signboards as much as possible. In response to such criticism, digital signage has attracted attention in recent years. In recent years, digital signage has tended to become larger and cheaper.

With digital signage, the displayed content can be changed as appropriate. A single information signboard can switch between a guide to a leisure facility and a plurality of commercial advertisements. As a result, no other signboards are needed.

By the way, it is presumed that the time period during which a driver heading to a leisure facility such as a golf course requires a guidance display is limited. For example, it is common to play golf from morning until early afternoon, and drivers heading to the golf course concentrate on the morning hours. Note that there is no need for a guide display for drivers returning home from the golf course.

By increasing the number of commercial advertising displays during times when there is little need for information displays for drivers, advertising revenue can be obtained and digital signage can be maintained and operated. Here, the time period and ratio of guidance display are important. If the guidance information is inadequate, drivers may get lost and may not be able to reach their destination. If the information is displayed excessively, sufficient advertising revenue cannot be obtained. Therefore, the prediction system of the present invention determines the optimal guidance display switching.

<Passing amount prediction and time zone setting>
FIG. 4 is an image diagram of the basic operation of the prediction system of the present application. FIG. 5 is an image diagram of each data.

First, it is estimated how many drivers need to visually check the guidance display, that is, the amount of demand. It would be good if the traffic volume heading towards the destination could be directly measured by checking the guide signboard, but it is difficult to extract only the vehicles heading towards the destination from the passing vehicles, so for example, estimation is performed as follows.

It is possible to install a sensor (for example, a camera) on the guide signboard 3 and actually measure the amount of traffic passing toward the destination (measured data A).

Furthermore, it is possible to install a sensor (for example, a camera) at the entrance of a destination and actually measure the amount of traffic arriving at the destination (actual measurement data B).

Actual measurement data A also includes vehicles heading toward the destination, but unrelated to the destination. Actual measurement data B also includes vehicles that arrived at their destinations via routes unrelated to the information signboard. Therefore, vehicles included in measured data A and measured data B are extracted while taking into account the time required from the guide sign to the destination. The vehicle may be identified by image analysis. Thereby, the amount of vehicles passing in front of the guide sign can be estimated based on actual measurements. By making this a time series, traffic volume distribution data (estimated data C) can be obtained.

Incidentally, it is generally assumed that the demand for leisure facilities such as golf courses is influenced by weather conditions (temperature, precipitation, wind speed, amount of solar radiation, etc.) and date and time information such as whether it is a holiday or not. For example, if it is a holiday and the weather is good, there is a high possibility that there will be many visitors. On the other hand, if it is a weekday or bad weather, there is a high possibility that there will be fewer visitors. In addition, people tend to stay at leisure facilities such as golf courses for a long time, and it is thought that the time when the vehicles of visitors to their destinations pass in front of the information signboards is concentrated in the morning. Identify other parameters that affect demand forecasting.

Analyze the correlation between past input parameters and past traffic distribution data (estimated data C) and build a prediction model. For example, a predictive model is generated using machine learning.

Furthermore, the parameters for the day are predicted and input into the prediction model to predict traffic distribution data (prediction data D). That is, the prediction data D is a demand prediction regarding drivers who require guidance display. Due to the unique nature of leisure facilities, there are times when demand is concentrated.

Note that the above-mentioned method of predicting the amount of passage is one example, and other methods may be used.

The guidance information display time zone setting unit 14 extracts a time zone where demand is concentrated in accordance with the prediction data D, and sets a time zone in which the guidance display is preferentially displayed. For example, during this time period, the ratio of guidance display is set to be 60% or more. Specifically, a guide is displayed for 10 seconds, then a commercial advertisement is displayed for 5 seconds, and this is repeated. Outside of this time period, the proportion of guidance display shall be 40% or less. Specifically, a guide is displayed for 5 seconds, then a commercial advertisement is displayed for 10 seconds, and this is repeated.

In the above example, to simplify the explanation, the time period is divided into two: the time period during which the information display is given priority and the other time periods. However, the time period may be further subdivided and the proportion of information display set according to demand in accordance with the predicted data D.

In the above example, as a specific example of giving priority to guidance display by the guidance information display time zone setting unit 14, attention was focused on the display time, but attention may be focused on the display area. For example, a guide display and a commercial advertisement display guide display are displayed on the same screen. In the guidance display priority time period, the area ratio of the guidance display is set to be 70% or more, and the area ratio of the commercial advertisement display is set to be less than 30%. Outside of this time period, the area ratio of guidance display shall be less than 30%, and the area ratio of commercial advertisement display shall be 70% or more.

In addition, the disadvantages caused by drivers overlooking guidance displays (for example, the number of complaints from drivers) are evaluated as profit reduction, and reinforcement learning is performed to maximize advertising revenue including the profit reduction evaluation, and the time period for guidance display is set. You may. In other words, the number of commercial advertisements displayed will be increased to the extent that guidance displays can function properly.

<Correction>
Traffic volume distribution data (prediction data D) is a prediction based on past measured data. It is not guaranteed that the trends of past data will be reproduced exactly as they are. In addition, although it is assumed that there is a correlation between the input parameters and the prediction target (traffic volume distribution), the direct causal relationship is unclear. Therefore, there is a limit to the prediction accuracy. Therefore, the traffic volume distribution data (prediction data D) is corrected based on the predicted passage time obtained from the navigation system.

A visitor (driver or navigation terminal user) to a leisure facility such as a golf course inputs a destination (leisure facility) into the navigation terminal 2 when heading to the leisure facility from home or the like.

The passing time prediction unit 13 predicts the time when the vehicle carrying the navigation terminal 2 will pass in front of the guide signboard. Note that if there is no guide signboard on the predicted route, the passage time prediction is omitted.

The predicted passage times are obtained from multiple (large number of) navigation terminals 2 and organized into a time series to generate predicted data E. The predicted passage times are updated as needed even while the vehicle is moving, and the predicted data E is updated as needed.

The predicted data E reflects only the information of visitors who have input their destinations into the navigation terminal 2. In other words, the predicted data E does not reflect information about visitors who do not use the navigation system.

However, since most drivers nowadays use navigation systems, a large amount of information can be obtained. In addition, since most visitors who input their destination into the navigation terminal 2 head to that destination, the prediction reproducibility is high. Furthermore, the predicted passage time is predicted, for example, several hours into the future, and since the prediction accuracy of the navigation system is high, the prediction accuracy of the predicted passage time is also high. Therefore, it can be assumed that the predicted data E is close to or similar to the actual number of vehicles passing in front of the guide sign.

In other words, the prediction data D is based on correlation, but the prediction data E is based on causality.

The modification unit 17 modifies the traffic volume distribution data (prediction data D) based on the predicted passage time (prediction data E) by the passage time prediction unit 13, and generates the predicted passage time (prediction data F). Further, the time slot for preferentially displaying the guidance information set by the guidance information display time slot setting unit 14 is modified. This improves prediction accuracy.

<Output and prediction model update>
The output unit 15 outputs display switching information (switching command between guidance display and commercial advertisement display) to the guidance signboard 3 via the communication unit 11. Based on the switching command, the guide signboard 3 gives priority to guidance display or displays priority to commercial advertisement display. Drivers passing in front of the information sign visually check the information display.

This will allow them to attract drivers reliably and increase advertising revenue.

The prediction system of the present invention not only predicts the amount of passage, but also measures the amount of passage.

The prediction model updating unit 16 readjusts the prediction model so that the corrected prediction data F approaches estimated data C (quasi-actual measurement data) that is based on actual measurement. This also improves the prediction accuracy of the prediction model.

~ Information signboard display switching modification example ~
<Summary>
FIG. 6 is an image diagram of a modified example of switching the guide signboard display.

The above application example and this modification have in common that the number of billboards is reduced by using digital signage, and that the optimal guidance display switching is determined.

In the application example described above, the prediction model generated by machine learning is corrected using the predicted time by the navigation system, whereas in this modification, the optimal guidance display switching is determined based on the predicted time by the navigation system.

The information processing device 1 and the plurality of navigation terminals 2 can communicate with each other via a network. Wide area wireless communication such as a mobile phone network is used for communication. However, the locations where information signs for leisure facilities such as golf courses are installed are often in mountainous areas where wide-area wireless communication is not possible. There is also the risk of problems with guidance provided by the navigation system, so visual recognition of guide signs is important.

Therefore, in this modification, a base station for short-range wireless communication is provided on the guide signboard or near the guide signboard. The short-range wireless communication base station can communicate with the information processing device 1 by wire or wirelessly. Note that the navigation terminal 2 is capable of both wide-area wireless communication and short-range wireless communication. Therefore, the navigation terminal 2 can also communicate with the short-range wireless communication base station.

It is difficult to clearly distinguish between wide-area wireless communication and short-range wireless communication, but for the sake of simplicity, wide-area wireless communication refers to communication distances of 100 m or more (generally several km), and communication distances of less than 100 m (generally several km) are defined as wide-area wireless communication. and 10 meters or more) is considered short-range wireless communication.

<Switching command>
A visitor (driver or navigation terminal user) to a leisure facility such as a golf course inputs a destination (leisure facility) into the navigation terminal 2 when heading to the leisure facility from home or the like.

The terminal information acquisition unit 12 acquires terminal information including current position information of the terminal, input destination information, and terminal identification information from the navigation terminal 2 via the communication unit 11 .

The passing time prediction unit 13 predicts the time when the vehicle carrying the navigation terminal 2 will pass in front of the guide signboard. Note that if there is no guide signboard on the predicted route, passing time prediction is omitted. The current position of the navigation terminal 2 is recognized even while moving, and the predicted passing time is updated at any time.

Immediately before the predicted passing time, the navigation terminal identification unit 18 (not shown) checks and identifies the terminal identification information of the navigation terminal 2 via the short-range wireless communication base station. For example, the terminal identification information of the navigation terminal 2 is acquired again, and it is determined whether it matches the terminal identification information of the navigation terminal 2 that was acquired last. If they match, the navigation terminal 2 is specified.

The switching command output unit 15 outputs display switching information (switching command between guidance display and commercial advertisement display) to the guidance signboard via the short-range wireless communication base station. Specifically, when the navigation terminal 2 is identified, a command is given to switch to the guidance display. The guidance signboard is basically used to display commercial advertisements, and will display the guidance display upon receiving a switching command. Drivers passing in front of the information sign visually check the information display. The information signboard returns to commercial advertisement display after a predetermined period of time has elapsed.

This makes it possible to reliably guide drivers and increase advertising revenue.

～Guidance notification modification example～
<Summary>
FIG. 7 is an image diagram related to the guidance notification.

This modification is a further modification of the above guide signboard display switching. In recent years, there has been a demand for the removal of information signs for safety and aesthetic reasons.

The information processing device 1 and the plurality of navigation terminals 2 can communicate with each other via a network. Wide area wireless communication such as a mobile phone network is used for communication. However, the locations where information signs for leisure facilities such as golf courses are installed are often in mountainous areas where wide-area wireless communication is not possible. There is also the risk of problems with guidance provided by the navigation system, so visual recognition of guide signs is important. In this modification, the guide signboard is removed and the navigation terminal 2 is notified of the guide display instead.

<Information notice>
A base station for short-range wireless communication will be installed at or near the guidance removal location. The short-range wireless communication base station can communicate with the information processing device 1 by wire or wirelessly. Note that the navigation terminal 2 is capable of both wide-area wireless communication and short-range wireless communication. Therefore, the navigation terminal 2 can also communicate with the short-range wireless communication base station.

A visitor (driver or navigation terminal user) to a leisure facility such as a golf course inputs a destination (leisure facility) into the navigation terminal 2 when heading to the leisure facility from home or the like.

The terminal information acquisition unit 12 acquires terminal information including current position information of the terminal, input destination information, and terminal identification information from the navigation terminal 2 via the communication unit 11 .

The passing time prediction unit 13 predicts the time when the vehicle carrying the navigation terminal 2 will pass the guidance removal position. Note that if there is no guide removal position on the predicted route, passing time prediction is omitted. The current position of the navigation terminal 2 is recognized even while moving, and the predicted passage time is updated at any time.

Immediately before the predicted passing time, the navigation terminal identification unit 18 (not shown) checks and identifies the terminal identification information of the navigation terminal 2 via the short-range wireless communication base station. For example, the terminal identification information of the navigation terminal 2 is acquired again, and it is determined whether it matches the terminal identification information of the navigation terminal 2 that was acquired last. If they match, the navigation terminal 2 is specified.

The guidance notification output unit 15 outputs a guidance notification to the specified navigation terminal 2 via the short-range wireless communication base station. Even if a problem occurs in the guidance provided by the navigation system due to the inability of wide area wireless communication, guidance will be displayed on the navigation terminal 2. This allows the driver to be reliably guided.

~Restaurant cooking schedule~
＜Overview＞
FIG. 8 is an image diagram of an example in which the prediction system of the present invention is applied to the cooking schedule of a large restaurant.

Traditionally, in large restaurants, food is mostly prepared in a central kitchen set up in a separate location, then the processed food is frozen and transported to the store, where simple finishing touches such as defrosting are performed. Serves food. On the other hand, in recent years, so-called local production for local consumption, which means eating local specialty products locally, has been attracting attention. With local production for local consumption, you can enjoy fresh local ingredients.

Whereas with the central kitchen system, cooking schedules can be created in the same way as for industrial products, with the local production and consumption system, both demand and supply are unstable, making it difficult to create cooking schedules. In particular, if the ingredients are frozen or preservatives are used, the advantage of fresh local ingredients is lost. In other words, it is not possible to prepare food in advance. In this respect, too, it is difficult to create cooking schedules with the local production and consumption system. Therefore, the present prediction system creates cooking schedules for restaurants that use the local production and consumption system.

<Creating a cooking schedule>
FIG. 9 is an image diagram of the cooking schedule.

In general, it is assumed that demand at restaurants is influenced by weather conditions (temperature, precipitation, wind speed, amount of sunlight, etc.) and date and time information such as whether it is a holiday or not. For example, if it is a holiday and the weather is good, there is a high possibility that there will be many visitors. On the other hand, if it is a weekday or bad weather, there is a high possibility that there will be fewer visitors. Additionally, demand is concentrated during lunch and dinner hours. Identify other parameters that affect demand forecasting. For example, seasonal ingredients vary depending on spring, summer, fall, and winter.

A predictive model is constructed by analyzing the correlation between past input parameters and past amounts of each ingredient used and amount of each dish served (including time series). For example, a predictive model is generated using machine learning.

The cooking schedule creation unit 14 predicts the parameters for the day, inputs them into a prediction model, and predicts the amount of each ingredient to be used and the amount of each dish to be served (including time series). Create a cooking schedule that reflects the compatible ingredients and cooking time used for each dish.

In the illustrated example, in order to simplify the explanation, the process is divided into preparation of each ingredient used and main cooking of each served dish. The time for pre-processing each ingredient is set so as to correspond to the demand time distribution for each served dish.

Note that the above demand forecasting method is just one example, and other methods may also be used.

The output unit 15 outputs the cooking schedule to a terminal 4 installed in the restaurant via the communication unit 11. The restaurant operator views the cooking schedule displayed on the terminal. The cooking schedule also shows the amount of each ingredient used. A rough cooking schedule can be obtained the day before. The restaurant operator orders and procures each ingredient by the day before so that it can be prepared according to the cooking schedule.

<Correction>
The above demand forecast is based on past actual measurement data. Trends in past data may not necessarily be reproduced exactly as they are. Furthermore, although it is assumed that there is a correlation between the input parameters and the prediction target (the amount of each dish served, etc.), the direct causal relationship is unknown. Therefore, there is a limit to prediction accuracy. Therefore, the demand forecast and cooking schedule are revised based on the predicted arrival time obtained from the navigation system.

A customer (driver or navigation terminal user) who intends to eat at a restaurant inputs a destination (a restaurant or a facility including a restaurant) into the navigation terminal 2 when heading to the restaurant from home or the like.

The arrival time prediction unit 13 predicts the time when the vehicle carrying the navigation terminal 2 will arrive at the restaurant.

By acquiring predicted arrival times from a plurality of (many) navigation terminals 2 and arranging them in chronological order, a distribution of predicted arrival times is generated. Even while the vehicle is moving, the predicted arrival time is updated at any time, and the expected store arrival time distribution is updated at any time.

Note that it is estimated that the time from when a vehicle arrives at the parking lot of a restaurant to when a customer is seated at the restaurant is approximately constant. That is, the store arrival time distribution occurs about a predetermined time (for example, 10 minutes) later than the arrival time distribution.

The expected store arrival time distribution reflects only information on visitors who have input their destinations into the navigation terminal 2. In other words, information about visitors who do not use the navigation system or who access by means other than cars is not reflected.

However, in recent years, most drivers use navigation systems, and many local restaurants are accessible by car, so a large amount of information can be obtained. Further, since most of the visitors who input their destination into the navigation terminal 2 head for the destination, the prediction reproducibility is high. Further, the predicted arrival time is predicted several hours in the future, and the navigation system has a high prediction accuracy, so the prediction accuracy of the predicted arrival time is high. Therefore, it can be inferred that the above expected store visit time distribution is approximate or similar to the actual store visit time distribution.

In other words, the amount of each dish to be served predicted up to the previous day is based on a correlation, but the distribution of expected arrival times on the day is based on a causal relationship.

The modification unit 17 modifies the prediction of the amount of each dish to be served and the cooking schedule based on the expected arrival time distribution obtained by the arrival time prediction unit 13. This improves prediction accuracy.

The output unit 15 outputs the day's cooking schedule to the terminal 4 installed in the restaurant via the communication unit 11. The restaurant operator views the cooking schedule displayed on the terminal.

Customers who arrive at a restaurant park their cars in the parking lot, come to the restaurant, and take a seat. Freshly prepared food is served to customers with almost no waiting time.

If the predicted amount of each ingredient to be used for the day is greater than the amount of each ingredient to be procured, the excess ingredients are processed to maintain their freshness and, if possible, used the next day. If the predicted amount of each ingredient to be used on the day is less than the amount of each ingredient to be procured, additional orders are placed for the missing ingredients on the day. Locally produced and locally consumed restaurants can easily procure ingredients in small quantities.

This prevents food from being unavailable due to running out of ingredients, and prevents excessive amounts of ingredients from being wasted.

<Prediction model update>
The prediction system of the present invention predicts the amount of each ingredient used and the amount of each dish served (time series), and also measures the amount of each ingredient used and the amount of each dish served (time series).

The predictive model updating unit 16 readjusts the predictive model so that the corrected predicted values of the amount of each ingredient to be used and the amount of each dish to be served approach the actual measured values. This also improves the prediction accuracy of the prediction model.

In addition, reinforcement learning is performed to minimize the profit reduction evaluation by converting the profit reduction evaluation due to the waiting time from the customer's order to the food being served, the profit reduction evaluation due to the shortage of ingredients, and the profit reduction evaluation due to food waste. You may do so.

<Modified example>
FIG. 10 is an image diagram of a modified example of cooking schedule creation.

The terminal information acquired by the terminal information acquisition unit 12 includes terminal identification information of the navigation terminal 2.

Preference information of the user of the navigation terminal 2 is stored in the database of the information processing device 1 or in the database of another information processing device capable of communicating with the information processing device 1 in association with each piece of terminal identification information. Therefore, the preference information is included in the terminal information acquired by the terminal information acquisition unit 12. The preference information acquisition unit is part of the terminal information acquisition unit 12.

It is presumed that there is a correlation between preference information and the food ordered at a restaurant. Customers are likely to order dishes associated with preference information. The arrival time prediction unit 13 generates an expected arrival time distribution for each offered dish, in other words, an expected order time distribution.

The modification unit 17 modifies the prediction of the amount of each dish to be served and the cooking schedule based on the expected order time distribution. Since it is possible to predict even the dishes that will be ordered, prediction accuracy is further improved.

～Parking situation prediction～
<Summary>
FIG. 11 is an image diagram of an example in which the prediction system of the present application is applied to predicting parking conditions in parking lots of leisure facilities such as amusement parks.

Amusement parks and other leisure facilities have the capacity to attract thousands to tens of thousands of people. Parking lots capable of accommodating this number are required. If one parking lot is not sufficient, several parking lots are set up around the leisure facility.

When parking lots are set up over a wide area, it is possible that some areas are congested and some have vacant spaces. However, visitors (drivers) are unable to grasp the parking situation at the parking lot, and may end up waiting for a parking spot even when there are other spaces available. Even if the parking lot manager attempts to guide drivers to an area with vacant spaces, the drivers may become confused because they are unable to grasp the overall situation. Therefore, the present prediction system predicts the parking situation at the parking lot and notifies drivers in advance.

<Parking situation prediction>
FIG. 12 is an image diagram of the parking situation prediction.

If there are many visitors, the parking lot will be crowded. If there are few visitors, the parking lot will be empty. Therefore, the parking situation can be predicted based on visitor demand.

In general, it is assumed that the demand for visitors to leisure facilities is influenced by weather conditions (temperature, precipitation, wind speed, amount of sunlight, etc.) and date and time information such as whether it is a holiday or not. For example, if it is a holiday and the weather is good, there is a high possibility that there will be many visitors. On the other hand, if it is a weekday or bad weather, there is a high possibility that there will be fewer visitors.

People tend to stay at leisure facilities for a long time, and arrival times are likely to be concentrated in the morning. On the other hand, if there is a nighttime event, the arrival time distribution will also change. The degree of congestion also changes depending on each visitor's route to the venue, such as when there are multiple parking lots or multiple entrances.

On the other hand, after a certain amount of staying time has passed, the parking lot becomes vacant. It is assumed that the length of stay will depend on weather conditions and the content of the event. Identify other parameters that affect demand forecasting.

Analyze the correlation between past input parameters and past parking conditions and build a predictive model. For example, a predictive model is generated using machine learning.

The parking situation prediction unit 14 predicts the parameters for the current day, inputs them into a prediction model, and predicts the parking situation of each parking lot at each time.

In the illustrated example, at the prediction target time, there is sufficient parking space in parking lot A, there is not enough space in parking lot B, and parking lot C is full. , it is expected that there will be some waiting time.

The output unit 15 publishes the predicted situation on a homepage or the like. The visitors view the homepage etc. in advance on their respective terminals 5, understand the predicted parking situation at the leisure facility, and select an available parking lot. Since we know the situation in advance, there is no confusion on site.

<Correction>
The above demand forecast is based on past measured data. It is not always possible to reproduce the trends of past data as is. In addition, although it is assumed that there is a correlation between the input parameters and the prediction target (parking situation prediction based on demand forecast), the direct causal relationship is unclear. Therefore, there is a limit to the prediction accuracy. Therefore, the demand forecast and parking situation forecast are revised based on the predicted arrival time obtained from the navigation system.

When a visitor to a leisure facility (a driver or a navigation terminal user) heads to the leisure facility from his/her home or the like, he/she inputs the destination (leisure facility) into the navigation terminal 2.

The arrival time prediction unit 13 predicts the time when the vehicle carrying the navigation terminal 2 will arrive at the leisure facility.

By acquiring predicted arrival times from a plurality of (many) navigation terminals 2 and arranging them in time series, an expected arrival time distribution is generated. Even while the vehicle is moving, the predicted arrival time is updated as needed, and the predicted arrival time distribution is updated as needed.

The above estimated arrival time distribution reflects only the information of visitors who have entered their destinations into the navigation terminal 2. In other words, it does not reflect the information of visitors who do not use the navigation system.

However, in recent years, most drivers use navigation systems, which provides a wealth of information. In addition, since most of the visitors who input their destination into the navigation terminal 2 head to the destination, the prediction reproducibility is high. Furthermore, since the predicted arrival time is predicted several hours in the future, and the navigation system has high prediction accuracy, the prediction accuracy of the predicted arrival time is high. Therefore, it can be inferred that the predicted arrival time distribution by the arrival time prediction unit 13 is approximate or similar to the actual arrival time distribution.

In other words, the expected arrival time distribution predicted up to the previous day is based on a correlation, but the expected arrival time distribution on the current day is based on a causal relationship.

The modification unit 17 modifies the parking situation prediction based on the expected arrival time distribution in the arrival time prediction unit 13. This improves prediction accuracy.

The output unit 15 updates the predicted situation display on the homepage or the like displayed on each user's terminal 5. Furthermore, it outputs guidance to the navigation terminal 2 via the communication unit 11. The driver is guided to an available parking lot by the guidance. Even if the driver does not understand the overall situation, he or she will not get confused on-site.

The navigation system takes into account the route and parking lot situation, selects the most suitable parking lot, and makes a suggestion to the driver. Drivers do not needlessly wait for a parking lot even if there is an empty parking lot.

<Prediction model update>
The prediction system of the present invention predicts the parking lot situation at each time and also measures the parking lot situation at each time.

The prediction model updating unit 16 readjusts the prediction model so that the corrected parking lot situation prediction approaches the actually measured parking lot situation. This also improves the prediction accuracy of the prediction model.

Furthermore, reinforcement learning can be used to minimize waiting times for parking spaces.

～Guidance staff placement plan～
<Summary>
FIG. 13 is an image diagram of an example in which the prediction system of the present invention is applied to a guide staff placement plan for a parking lot of a leisure facility such as an amusement park.

Amusement parks and other leisure facilities have the ability to attract thousands to tens of thousands of people. A large parking lot that can accommodate this will be required. If one parking lot is insufficient, several parking lots will be provided around the leisure facility.

When a parking lot becomes large-scale or is spread over a wide area, there will be a mixture of crowded areas and vacant areas. Additionally, there is a time lag when the parking lot becomes full. As a result, the need for parking lot attendants varies.

In recent years, there has been a trend of labor shortage. It is important to make effective use of limited human resources. Focus on placing guides in locations and times when they are most needed. Therefore, the prediction system of the present application creates a guide staff allocation plan.

<Guidance staff placement plan>
FIG. 14 is an image diagram of the guide staff allocation plan.

When there are many visitors, the parking lot becomes crowded and the need for guides increases. When there are few visitors, the parking lot will be empty and the need for guides will not be as high. Therefore, visitor demand is an important factor when formulating a guide staff allocation plan.

In general, it is assumed that visitor demand at leisure facilities is influenced by weather conditions (temperature, precipitation, wind speed, solar radiation, etc.) and date and time information such as whether it is a holiday or not. For example, if it is a holiday and the weather is good, there is a high probability that there will be many visitors. Conversely, if it is a weekday or the weather is bad, there is a high probability that there will be few visitors.

People tend to stay at leisure facilities for a long time, and arrival times are likely to be concentrated in the morning. On the other hand, if there is a nighttime event, the arrival time distribution will also change. The degree of congestion also changes depending on each visitor's route to the venue, such as when there are multiple parking lots or multiple entrances.

On the other hand, after a certain amount of time has passed, people will start heading home. The length of time people stay is presumably affected by weather conditions and the content of the event. When there are a large number of people heading home, the need for guides increases.

Identify other parameters that influence the need for ushers.

Analyze the correlation between past input parameters and the past need for guides and guide deployment situations to build a predictive model. For example, generate a predictive model using machine learning.

The guide placement planning unit 14 predicts parameters for the day, inputs them into a prediction model, predicts the need for guides at each time, and deploys guides to locations where the need for guides is high.

In the illustrated example, initially, one guide person is placed in each of Parking Lot A, Parking Lot B, and Parking Lot C, and it is predicted that parking lot A will be crowded in the order of parking lot B → parking lot C. The number of guides has been reinforced accordingly. In particular, we predict that by noon, Parking Lot A will be full and the need for guides will be reduced, while Parking Lot C will be crowded with visitors and the need for guides will increase. Parking lot C will have no guide staff, and three guide staff will be stationed in parking lot C.

In anticipation that visitors will begin returning home from Parking Lot A in the afternoon, two guides will be stationed in Parking Lot A. From then on, the number of guides will be prioritized depending on the congestion of returning passengers.

The output unit 15 outputs the guide placement plan to the parking lot manager's terminal 4 via the communication unit 11. The parking lot manager views the guide placement plan displayed on the terminal. A rough guide staff placement plan can be obtained the day before. The parking lot manager verifies whether there are any defects in the guide staff placement plan based on the prediction system of the present application and approves it.

The output unit 15 outputs the schedule for each guide to the terminal 5 of each guide via the communication unit 11. Each guide works at his/her assigned location according to his/her own schedule.

<Correction>
The above demand forecast is based on past measured data. It is not guaranteed that the trends of past data will be reproduced exactly as they are. In addition, although it is assumed that there is a correlation between the input parameters and the forecast target (the forecast of the need for guides based on the demand forecast), the direct causal relationship is unclear. Therefore, there is a limit to the accuracy of the forecast. Therefore, the demand forecast and the guide deployment plan are revised based on the predicted arrival time obtained from the navigation system.

A visitor to a leisure facility (driver or navigation terminal user) inputs a destination (leisure facility) into the navigation terminal 2 when heading to the leisure facility from home or the like.

The arrival time prediction unit 13 predicts the time when the vehicle carrying the navigation terminal 2 will arrive at the leisure facility.

A predicted arrival time distribution is generated by acquiring predicted arrival times from a plurality of (many) navigation terminals 2 and arranging them in a time series. Even while the vehicle is moving, the predicted arrival time is updated as needed, and the predicted arrival time distribution is updated as needed.

The above expected arrival time distribution reflects only the information of visitors who have input their destinations into the navigation terminal 2. In other words, information about visitors who do not use the navigation system is not reflected.

However, in recent years, most drivers use navigation systems, which provides a wealth of information. Moreover, since most of the visitors who input their destination into the navigation terminal 2 head for the destination, the prediction reproducibility is high. Further, the predicted arrival time is predicted several hours in the future, and the navigation system has a high prediction accuracy, so the prediction accuracy of the predicted arrival time is high. Therefore, it can be inferred that the predicted arrival time distribution by the arrival time prediction unit 13 is approximate or similar to the actual arrival time distribution.

In other words, the expected arrival time distribution predicted up to the previous day is based on a correlation, but the expected arrival time distribution on the current day is based on a causal relationship.

The correction unit 17 corrects the guide deployment plan based on the estimated arrival time distribution from the arrival time prediction unit 13. This improves the prediction accuracy.

The output unit 15 outputs and updates the guide staff allocation plan for the day to the parking lot manager's terminal 4 via the communication unit 11. The parking lot manager views, verifies, and approves the day's guide placement plan displayed on the terminal.

The output unit 15 outputs and updates the schedule for each guide to the terminal 5 of each guide via the communication unit 11. Each guide will work at their assigned location according to the change instructions.

<Prediction model update>
The prediction system of the present application not only predicts the need for guides, but also measures the need for guides. For example, the traffic volume of arriving and exiting vehicles and the residence time of vehicles waiting to be parked and vehicles waiting to exit are measured and used as indicators of the need for guides.

The prediction model updating unit 16 readjusts the prediction model so that the revised prediction of the need for the guide becomes closer to the actual measurement of the need for the guide. This also improves the prediction accuracy of the prediction model.

~Main point~
FIG. 15 is an image diagram of the main points of the present application.

This application consists of a demand forecasting system and a navigation system. A demand forecasting system builds a forecasting model based on the correlation of past data. The above prediction model is based on the correlation of past data, but since the direct cause-and-effect relationship is unknown, there is a limit to the prediction accuracy.

On the other hand, when a destination is input into a navigation system, the vehicle will almost always head to the destination. Also, the accuracy of the predicted arrival time is very high. Therefore, there is a high probability that the user of the navigation terminal 2 will arrive at the destination at the predicted arrival time. In other words, there is a high direct causal relationship compared to other parameters. In other words, it is comparable to actual measurement data.

On the other hand, the predicted arrival time based on the navigation system cannot be obtained the day before. Therefore, a rough demand forecast is made by the demand forecasting system by the previous day, and the demand is responded to and corrected on the day based on the expected arrival time distribution. This improves prediction accuracy.

There is a limit to the accuracy of predictions made up to the previous day, and even if appropriate measures are taken the day before, problems may occur on the day. The revised forecast is based on information from a few hours ago (strictly speaking, it is a forecast based on past data), but if there is a problem, it can be fixed within the next few hours (travel time from entering the destination to arriving at the destination). . In other words, the correspondence can be modified.

The present invention is characterized in that it focuses on making effective use of these several hours.

1 Information processing device 2 Navigation terminal 3 Display device (information signboard)
4 Management terminal 5 Mobile terminal 11 Communication unit 12 Terminal information acquisition unit 13 Arrival time prediction unit 14 Specific item prediction unit 15 Prediction output unit 16 Prediction model update unit 17 Modification unit 18 Navigation terminal identification unit

Claims (6)

a communication section capable of communicating with multiple navigation terminals;
a terminal input information acquisition unit that acquires, from the navigation terminal, terminal input information including a destination input to the navigation terminal, a current position at the time of input, and a current time at the time of input;
an arrival time prediction unit that predicts a time when the navigation terminal will arrive at a specific location including the destination based on information including the current location and the current time;
a specific matter prediction unit that applies a prediction model generated by machine learning to predict a specific matter at the specific location;
Generate arrival time prediction data based on the arrival time of the navigation terminal predicted by the arrival time prediction unit, and generate prediction information regarding the specific item performed by the specific item prediction unit based on the arrival time prediction data. a correction section to correct;
an output unit that outputs prediction information related to the specific matter corrected by the correction unit;
An information processing device comprising: A device that instructs the display device to switch at a specific position where a display device capable of switching and displaying guidance information to the destination and information other than the guidance information is provided, the device comprising:
The arrival time prediction unit predicts a time when the navigation terminal passes a specific position,
The specific item prediction unit applies a traffic volume estimation model generated by machine learning and sets a time period for displaying the guidance information in accordance with a traffic volume time distribution from the specific location to the destination. ,
The modification unit generates passing time prediction data based on the passage time of the navigation terminal predicted by the specific item prediction unit, and the guidance set by the specific item prediction unit based on the passage time prediction data. Modify the time period for displaying information,
2. The information processing device according to claim 1, wherein the output unit outputs a switching command to the display device so as to display the guidance information during the time period corrected by the correction unit. A device for creating a cooking schedule for food to be provided at the destination,
The arrival time prediction unit predicts a time when the navigation terminal will arrive at the destination,
The specific item prediction unit applies a prediction model generated by machine learning to create a cooking schedule for food to be provided at the destination,
The modification unit generates arrival time prediction data based on the arrival time of the navigation terminal predicted by the arrival time prediction unit, and the cooking schedule created by the specific item prediction unit based on the arrival time prediction data. Modify and
2. The information processing device according to claim 1, wherein the output unit outputs the cooking schedule modified by the modification unit to a terminal installed at a food serving location. the terminal input information includes a terminal ID of the navigation terminal,
A preference information acquisition unit that acquires preference information stored for each terminal ID,
The information processing apparatus according to claim 3 , wherein the correction unit corrects the cooking schedule created by the specific item prediction unit based on the arrival time of the navigation terminal predicted by the arrival time prediction unit and the preference information. A device that predicts parking conditions in parking lots around the destination,
The arrival time prediction unit predicts the time when the navigation terminal will arrive at the destination,
The specific item prediction unit applies a prediction model generated by machine learning to predict the parking situation of a parking lot around the destination,
The modification unit generates arrival time prediction data based on the arrival time of the navigation terminal predicted by the arrival time prediction unit, and the parking situation created by the specific item prediction unit based on the arrival time prediction data. Modify forecast information,
2. The information processing device according to claim 1, wherein the output unit outputs the parking situation prediction information and guidance corrected by the correction unit to the navigation terminal. A device for managing the placement of guides in parking lots around the destination, the device comprising:
The arrival time prediction unit predicts the time when the navigation terminal will arrive at the destination,
The specific matter prediction unit plans the placement of guides to parking lots around the destination by applying a prediction model generated by machine learning,
The modification unit generates arrival time prediction data based on the arrival time of the navigation terminal predicted by the arrival time prediction unit, and adjusts the guide planned by the specific item prediction unit based on the arrival time prediction data. Modify the placement plan information,
2. The information processing device according to claim 1, wherein the output unit outputs the guide staff placement plan information modified by the modification unit to a terminal of a parking lot manager.