A prediction device, prediction method, and program for simulating a vehicle dispatch system.

Abstract

We predict performance indicators for each vehicle allocation in the dispatch system. [Solution] In the prediction device 101, the prediction unit 102 predicts the performance indicators of the dispatch system for each of the multiple vehicle allocations by performing a simulation based on at least the following: the user characteristics related to the difference between the departure point and the pickup point in the itinerary, the difference between the arrival point and the drop-off point in the itinerary, the expected waiting time from the start of the itinerary to the pickup, the expected duration of the itinerary, the difference between users and the difference between the departure point and the pickup point in the itinerary, the difference between the arrival point and the drop-off point in the itinerary, the expected waiting time from the start of the itinerary to the pickup, the expected duration of the itinerary, and the traffic characteristics related to the traffic in the area to which the multiple vehicles and the dispatch system are applied. The illustration unit 103 illustrates the proportion of multiple constraints in each vehicle allocation and the predicted performance indicators for each vehicle allocation.

Description

Technical Field

[0008] , ,

[0009] , ,

[0001] The present invention relates to a prediction device, a prediction method, and a program for predicting performance indicators of a vehicle allocation system for each vehicle allocation in which constraints that a pickup point and a drop-off point should satisfy when a user is picked up, transferred, and then dropped off are set for a plurality of vehicles by simulating the vehicle allocation system.

Background Art

[0002] Conventionally, a technique for planning a route when allocating shared vehicles has been proposed (for example, Patent Document 1).

[0003] In such a vehicle allocation system, its performance may be examined based on the following performance indicators.

[0004] (a) Cancellation rate, which represents the ratio of users who cancel the vehicle allocation after seeing the predicted waiting time of the vehicle allocation and the predicted travel time of the travel requested by the user to the travel requested by the user.

[0005] (b) Predicted waiting time of the vehicle allocation that the vehicle allocation system responds to for the travel requested by the user.

[0006] (c) How much the actual waiting time after vehicle allocation exceeds the predicted waiting time predicted before vehicle allocation.

[0007] (d) How much the actual travel time required exceeds the predicted travel time predicted before vehicle allocation.

[0008] Regarding the performance indicators related to the above (a)-(d), the smaller each of them is, the better the performance of the vehicle allocation system.

[0009] On the other hand, regarding the pickup point for picking up a user and the drop-off point for dropping off a user, different constraints can be imposed for each shared vehicle as follows.

[0010] (1) Urban constraints for urban SAVs (Smart Access Vehicles) that are designed to transport passengers within urban areas by picking them up at one of several pre-set points within the urban area and dropping them off at another point. Urban SAVs are sometimes called CSAVs (City SAVs).

[0011] (2) Suburban spot constraints for a Suburban Spot SAV that is dedicated to transporting passengers between urban and suburban areas by picking up passengers at a relatively small number of pre-determined suburban spot locations such as tourist spots in the suburbs and dropping them off at a large number of pre-determined urban spots within the urban area. A Suburban Spot SAV is sometimes called an SSAV (Suburban SAV).

[0012] (3) All-area constraints for all-area SAVs that allow users to be picked up and dropped off at either suburban or urban locations. All-area SAVs are sometimes called ASAVs (All-area SAVs).

[0013] Furthermore, outside of areas where only passenger transport vehicles are in operation, passengers may use other public transportation such as buses and taxis.

[0014] Buses are subject to geographical constraints, such as picking up and dropping off passengers at predetermined bus stops, and these constraints are generally considered stricter than those of public transport. Furthermore, passengers need to walk from their starting point to the pickup point and from the drop-off point to their destination. Also, because buses operate on a predetermined schedule, waiting times for pickup (including walking time from the starting point to the pickup point) tend to be long.

[0015] Taxis pick up passengers at their desired departure point and drop them off at their desired destination, so the departure point and pickup point coincide, and the arrival point and drop-off point also coincide. Therefore, the geographical constraints on taxi operations are considered to be less restrictive than those on buses.

[0016] For example, even in the case of shared vehicles subject to urban area constraints, there are forms in which passengers are picked up at their desired departure point and dropped off at their desired destination, or where the pickup and drop-off points are limited to predetermined major roads. Furthermore, in the case of suburban spot constraints, there are forms in which drop-off points in suburban areas are limited to designated shared ride pools. Therefore, the geographical constraints on shared vehicles can be considered to be somewhere between those of buses and taxis.

[0017] If the waiting time or walking time is long when using the cheapest bus, users will consider using a taxi or shared ride. When comparing taxis and shared rides, they will decide which to use by taking into account the increased travel time due to other passengers, the possibility of delays in arrival, and the amount of cost involved. [Prior art documents] [Patent Documents]

[0018] [Patent Document 1] Patent No. 7294660 [Overview of the project] [Problems that the invention aims to solve]

[0019] Ride-hailing systems often have to consider not only shared taxis but also other forms of public transportation such as buses and taxis. Therefore, they need to be able to accommodate situations with varying geographical constraints.

[0020] While existing buses and taxis have a fixed number of vehicles, and therefore the number of vehicles allocated to geographical constraints is also fixed, shared transportation vehicles offer more flexibility in allocating the number of vehicles subject to urban area constraints, suburban spot constraints, area-wide constraints, and other constraints.

[0021] Therefore, there is a desire to improve the performance of the dispatch system by adjusting the proportion of restrictions imposed when allocating the number of SAVs (Service Vehicles) for different types of shared transportation, such as how many SAVs should be for urban areas, how many for suburban spots, how many for the entire area, and how many for other new types of SAVs. Furthermore, there are also cases where adjustments to the allocation of vehicles other than shared transportation (including buses, taxis, etc.) are desired.

[0022] To achieve this, it is necessary to predict the performance indicators of the dispatch system based on the proportion of constraints in vehicle allocation.

[0023] The present invention aims to solve the above problems by providing a prediction device, prediction method, and program that predict the performance indicators of the vehicle allocation system for each vehicle allocation, which sets constraints that the pickup point and drop-off point must satisfy when picking up users, transporting them, and then dropping them off, by simulating the vehicle allocation system for each of the vehicle allocations. [Means for solving the problem]

[0024] The prediction device according to the present invention is a prediction device that simulates a dispatch system in which multiple geographical constraints that must be met by the pickup point and drop-off point when multiple vehicles pick up and transport users and then drop them off are prepared in advance, and multiple vehicle allocations are prepared in advance in which one of the constraints from the multiple constraints is set for each of the multiple vehicles, For each of the above-mentioned allocations of the number of units, When a plurality of users who intend to use the vehicle dispatching system during a journey moving from a departure point to an arrival point make selections regarding the availability of use according to the differences between the departure point and the pick-up point in the journey, the differences between the arrival point and the drop-off point in the journey, the expected waiting time from the start of the journey to pick-up, and the expected time required for the journey, user characteristics related thereto, and traffic characteristics related to the traffic in the area to which the plurality of vehicles and the vehicle dispatching system are applied, by performing the simulation based at least on these, predict the performance index of the vehicle dispatching system for each allocation of the number of vehicles, and illustrate the ratio of the plurality of constraints in each allocation of the number of vehicles and the predicted performance index for each allocation of the number of vehicles.

Advantages of the Invention

[0025] According to the present invention, by performing a simulation of a vehicle dispatching system for each allocation of the number of vehicles that sets constraints to be satisfied by pick-up points and drop-off points when dropping off after picking up and transporting users, a prediction device, a prediction method, and a program for predicting the performance index of the vehicle dispatching system for each allocation of the number of vehicles can be provided.

Brief Description of the Drawings

[0026] [Figure 1] It is an explanatory diagram showing the configuration of a prediction device according to an embodiment of the present invention. [Figure 2] It is a flowchart showing the flow of control of prediction processing according to an embodiment of the present invention. [Figure 3] It is a triangular graph illustrating an example of the relationship between the ratio of constraints in each allocation of the number of vehicles and the predicted performance index for each allocation of the number of vehicles. [Figure 4] It is an explanatory diagram arranging triangular graphs illustrating the performance indexes predicted for combinations in which SAInterval and SAUrg are changed with respect to SAPos = 10, SAVN = 5, and CAInterval = 60. [Figure 5]This is an explanatory diagram showing a series of triangular graphs illustrating the predicted performance indicators for combinations of SAInterval and SAUrg, given SAPos=20, SAVN=5, and CAInterval=60. [Figure 6] This is an explanatory diagram showing a series of triangular graphs illustrating the predicted performance indicators for combinations of SAInterval and SAUrg, given SAPos=10, SAVN=6, and CAInterval=60. [Figure 7] This is an explanatory diagram showing a series of triangular graphs illustrating the predicted performance indicators for combinations of SAInterval and SAUrg, given SAPos=20, SAVN=6, and CAInterval=60. [Figure 8] This is an explanatory diagram showing a series of triangular graphs illustrating the predicted performance indicators for combinations of SAInterval and SAUrg, given SAPos=10, SAVN=6, and CAInterval=120. [Modes for carrying out the invention]

[0027] Embodiments of the present invention are described below. These embodiments are for illustrative purposes only and do not limit the scope of the present invention. Therefore, those skilled in the art can adopt embodiments in which each or all of the elements of these embodiments are replaced with equivalent elements. Furthermore, the elements described in each embodiment can be omitted as appropriate depending on the application. Thus, any embodiment configured according to the principles of the present invention falls within the scope of the present invention.

[0028] (composition) Figure 1 is an explanatory diagram showing the configuration of a prediction device according to an embodiment of the present invention. The following description will be based on this figure.

[0029] The prediction device 101 according to this embodiment is typically implemented by a computer executing a program. This computer is connected to various output and input devices and transmits and receives information with these devices.

[0030] Programs executed on a computer can be distributed and sold via a server to which the computer is connected for communication. In addition, they can also be recorded on non-transitory information storage media such as CD-ROMs (Compact Disk Read Only Memory), flash memory, and EEPROMs (Electrically Erasable Programmable ROMs), and then distributed and sold.

[0031] The program is installed on a non-temporary information storage medium such as a hard disk, solid-state drive, flash memory, or EEPROM on the computer. The computer then realizes the information processing device in this embodiment. Generally, the computer's CPU (Central Processing Unit), under the management of the computer's OS (Operating System), reads the program from the information storage medium into RAM (Random Access Memory), and then interprets and executes the code contained in the program. However, in architectures where the information storage medium can be mapped within the memory space accessible to the CPU, explicit loading of the program into RAM may be unnecessary. Furthermore, various information required during the program execution process can be temporarily stored in RAM.

[0032] Furthermore, it is desirable for computers to be equipped with a GPU (Graphics Processing Unit) in addition to a CPU for high-speed performance of various image processing calculations. By using a GPU and libraries such as TensorFlow, it becomes possible to utilize learning and classification functions in various artificial intelligence processes under the control of the CPU.

[0033] It is also possible to configure the information processing device of this embodiment using a dedicated electronic circuit instead of implementing it with a general-purpose computer. In this embodiment, the program can be used as material for generating wiring diagrams, timing charts, etc., of the electronic circuit. In this embodiment, an electronic circuit that satisfies the specifications defined in the program is configured using an FPGA (Field Programmable Gate Array) or ASIC (Application Specific Integrated Circuit), and this electronic circuit functions as a dedicated device that performs the functions defined in the program, thereby realizing the information processing device of this embodiment.

[0034] The prediction device 101 according to this embodiment is typically implemented by executing a program on various computers or by dedicated electronic circuits. For example, the prediction device 101 can be implemented on a computer with high computing speed specialized for parallel computing, such as a supercomputer or a high-performance computer.

[0035] In the following explanation, for the sake of clarity, the prediction device 101 will be described assuming that it is implemented by a computer executing a program.

[0036] The prediction device 101 according to this embodiment is Multiple vehicles are pre-defined with geographical constraints that must be met by the pickup and drop-off points when transporting and dropping off passengers. This simulation demonstrates a dispatch system where multiple vehicle allocations are pre-configured, each applying one of several constraints to multiple vehicles. It is a device that includes a prediction unit 102 and an illustration unit 103.

[0037] Here, the prediction unit 102, For each of the multiple allocations, Multiple users who intend to use a ride-hailing system for a journey from a departure point to an arrival point, the differences between the departure point and pickup point in the journey, the differences between the arrival point and drop-off point in the journey, the estimated waiting time from the start of the journey to pickup, and the user characteristics related to their decision on whether or not to use the system based on the estimated duration of the journey, Traffic characteristics related to traffic in areas where multiple vehicles and dispatch systems are applied, By conducting simulations based at least on this, we predict the performance indicators of the dispatch system for each vehicle allocation.

[0038] In this simulation, We accept requests from simulated users specifying at least the departure and arrival points. From among multiple vehicles, identify a vehicle capable of transporting a simulated user from a pickup point, which is a predetermined vicinity of the departure point, to a drop-off point, which is a predetermined vicinity of the arrival point. Based on traffic characteristics, the estimated waiting time from when a ride request is received until the vehicle arrives at the pickup location, and the estimated time from when the ride request is received until the vehicle arrives at the drop-off location or the simulated user arrives at the destination, are estimated. The system provides a response to the simulated user that includes at least the pickup and drop-off locations where the identified vehicle can transport the simulated user, the expected waiting time, and the expected total travel time. Based on the user characteristics related to the selection of the simulated user for whom a response was presented, a decision is made on whether to dispatch the vehicle related to the response or to cancel the dispatch. It is desirable to configure it in this way.

[0039] This embodiment considers various scenarios in which a (simulated) user utilizes the dispatch system. These various scenarios include: A situation where a person who wishes to travel by vehicle dispatched by a ride-hailing system operates an app or web service similar to a transit guidance app on their smartphone. The situation where hospitals use apps or web services for dispatching patients, Situations where schools use apps or web services for ride-hailing systems to transport students to and from school. Shopping centers using ride-hailing apps and web services for their customers. While various situations are conceivable and the present invention can be applied in such cases, the following description will assume the following mode of using the app as an example. In cases other than when the vehicle user themselves uses the app, for example, when hospital staff take a patient who has finished their examination to the nearest bus stop or train station, the present invention can also be applied.

[0040] In other words, with this app, the user specifies the departure point and the arrival point. They may also specify constraints on the departure and arrival times. The app then suggests one or more itineraries.

[0041] The app will present users with the following information for each itinerary: (a) The starting point or a nearby pickup point. The distance and estimated time required to travel on foot from the starting point to the pickup point. (b) The destination or a nearby drop-off point. The distance and estimated time required to travel on foot from the drop-off point to the destination. (c) Estimated waiting time from the start of the itinerary to pickup. (d) The estimated time required from the start to the end of the itinerary.

[0042] Furthermore, the following information can also be added: (e) The estimated cost of the entire itinerary. (f) The number of transfers in the itinerary and the estimated time required for each transfer. (g) Whether or not there are any planned shared rides during the itinerary. (h) Current location of vehicles heading to the pickup point, vehicles near the departure point, vehicles near the user's current location, etc.

[0043] The itinerary is as follows: This involves using only public transportation, such as buses and trains, which have strict restrictions on pickup and drop-off points, but are relatively inexpensive. This refers to using only transportation methods that have loose restrictions on pickup and drop-off locations, such as taxis, but are relatively expensive. This refers to transportation methods that only utilize intermediate means of transport, such as shared vehicles, which have restrictions on pickup and drop-off points and associated costs. These are combined as appropriate. However, the items are presented in the order specified by the user, or in the order recommended to the user.

[0044] The user then selects an itinerary within the app. If the itinerary requires a ride (for example, a taxi or shared vehicle), the app proceeds with the ride arrangement. If an itinerary that does not require a ride is selected, or if no itinerary is selected, the ride arrangement is canceled.

[0045] Reasons for canceling a ride include, from the user's perspective, long waiting times, high costs relative to the long waiting time, the overall journey taking too long, the existence of other modes of transportation with shorter waiting times that make a ride unnecessary (such as a bus), or too much time spent walking.

[0046] The characterization of such simulated users is represented by the user characteristics provided during the simulation. These user characteristics may be set based on the trends of existing users in the area assumed by the dispatch system, such as public transportation usage and surveys, or they may be used as is or modified as appropriate from user characteristics adopted in other areas.

[0047] Furthermore, if historical user statistics exist for the area where the ride-hailing system is applied, The number and percentage of users moving from section A to section B within the area during each time period. The urgency of the user's movement. This information can also be taken into account as user characteristics.

[0048] for example, Let Dist be the Manhattan distance between the pickup point and the drop-off point. Let V be the average speed of the vehicle. The estimated travel time for the vehicle in question is approximated by Dist / V. Note that any distance measure other than Manhattan distance (for example, Euclidean distance or the distance of the route within the map) can be used.

[0049] Here, the threshold for the duration of the itinerary that the user can tolerate is defined as Dist / (V×Urg) based on the urgency level Urg, which has a positive value.

[0050] Furthermore, user characteristics can be defined, such as continuing to dispatch a vehicle if the time obtained by adding a threshold to the departure time is earlier than the arrival deadline expected by the user, and canceling the dispatch if it is later.

[0051] For travel within urban areas, instead of using an urgency level, cancellation may occur if the waiting time exceeds a fixed threshold, or cancellation may occur with a certain probability.

[0052] Furthermore, traffic characteristics related to the areas where the ride-hailing system is applied include map and route information that predicts travel time using technology similar to car navigation systems, the days of the week, times of day, and severity of congestion, and the operating timetables used by public transportation.

[0053] Furthermore, information such as whether each section within the area to which the ride-hailing system is applied is an urban area, suburban area, tourist area, shopping district, government office, hospital, etc., may also be taken into account as part of the transportation characteristics.

[0054] In this configuration, as a performance indicator, (1) Cancellation rate of ride dispatches Predicting this is important when considering the performance of the unit allocation.

[0055] Furthermore, in configurations that include shared vehicles capable of transporting multiple passengers traveling on different itineraries, the following performance indicators can also be adopted. (2) Statistics based on the distribution of the estimated waiting time from the start of the journey to pickup, as presented to the simulated user for whom a shared vehicle was assigned (mean, mode, median, variance, minimum, maximum, etc.; the same applies hereinafter). (3) A statistical measure based on the distribution of the excess time of the simulated waiting time spent by a simulated user in the simulation, relative to the estimated waiting time from the start of the journey to pickup that was presented to the simulated user for whom a shared ride had been dispatched. For example, if a simulated user requests a ride in the simulation and is waiting for pickup when another ride request comes in, causing the vehicle to take a detour and arrive late at the pickup location, the actual waiting time will be longer than the estimated waiting time. This extended time corresponds to the "excess time" referred to here. (4) A statistical measure based on the distribution of the time exceeding the estimated travel time presented to the simulated user for the journey in the simulation, relative to the estimated travel time for the simulated user for whom a shared vehicle was dispatched. Similar to the time exceeding the estimated waiting time mentioned above, if other shared ride requests come in after a vehicle has been dispatched to the simulated user, the final arrival time will be delayed. This delayed time corresponds to the "excess time" referred to here.

[0056] Furthermore, the time and distance required to travel on foot from the drop-off point to the destination point can also be used as a criterion for whether or not a (simulated) user will cancel the ride, and statistical quantities based on the distribution of these factors can also be used as performance indicators.

[0057] On the other hand, the diagram section 103 illustrates the proportion of multiple constraints in each allocation of units and the predicted performance indicators for each allocation of units.

[0058] Here, if there are three types of multiple unit allocations and the performance indicators are cancellation rates or other scalar statistics, plotting them on a triangular graph with colors corresponding to the performance indicators makes it easier to analyze and compare the performance indicators of these multiple unit allocations.

[0059] (Predictive processing) Figure 2 is a flowchart showing the control flow of the prediction process according to an embodiment of the present invention. The following explanation will be given with reference to this figure.

[0060] In this process, the first step is to initialize the system by acquiring the following information related to the target of the prediction (step S301). Transportation characteristics of the target area. User characteristics of simulated users in the target area. Allocation of the number of vehicles to be included in the prediction.

[0061] For ease of understanding, this diagram limits the definition of user characteristics to one type. However, it is also possible to adopt a configuration where simulated users with different characteristics exist simultaneously, or to keep the characteristics of the simulated users constant in a single simulation and perform simulations for different characteristics separately, then calculate the average or weighted average of the obtained performance indicators.

[0062] Regarding the allocation of the number of units, For public transportation such as buses and taxis, the number of vehicles will be fixed or the distribution of vehicles during certain time periods will be fixed. For passenger vehicles subject to geographical constraints, combinations are generated with an upper limit on the total number of each vehicle type, such as ASAV, SSAV, and CSAV. Let's summarize each of these. This allows for the generation of multiple unit allocations that will be used for prediction.

[0063] Next, the prediction device 101 generates a list of requests related to the use of the dispatch system that will occur during the period to be simulated, based on the user characteristics (step S302). Each request generated here represents which simulated user wants to travel, when, and from where to where.

[0064] For ease of understanding, this diagram generates only one column of requirements. However, it is also possible to generate multiple columns, perform the simulation described below for each, and calculate the average or weighted average of the obtained performance metrics.

[0065] Next, the following process is repeated for each of the multiple allocations (step S303).

[0066] In other words, the above simulation is performed based on the generated request sequence, the acquired traffic characteristics, and the vehicle allocation to be processed in this iteration (step S304).

[0067] Then, the prediction device 101 calculates a performance index for the allocation of units based on the results of the simulation (step S305), records the result of associating the allocation of units with the calculated performance index (step S306), and repeats this process (step S307).

[0068] Once performance indicators have been recorded for all unit allocations, the prediction device 101 illustrates the unit allocations and their performance indicators using a triangular graph or the like based on the recorded results (step S308), and then terminates this process.

[0069] In step S306, information used to identify user characteristics, traffic characteristics, and request sequences used in the simulation may be recorded along with the vehicle allocation, and this information may be used as data, materials, raw data, and various decision-making criteria when conducting new simulations.

[0070] (Example display) Figure 3 is a triangular graph illustrating an example of the relationship between the proportion of constraints in each allocation and the predicted performance indicators for each allocation. The following explanation will refer to this figure.

[0071] The triangular graph shown in this figure illustrates the results of a simulation predicting the cancellation rate for a certain region using the following parameters. This simulation uses the Manhattan distance as the distance scale.

[0072] The users are divided into two groups: those who travel between urban and suburban areas separated by a distance of SAPos, and those who travel within urban areas defined by the distance citySize. citySize = 3km. SAPos = 10km. CAInterval = 60 ... The average time interval between randomly generated requests for movement within an urban area is 60 seconds. SAInterval = (90, 120) ... The average time interval between randomly generated requests to move between urban and suburban areas is 90 or 120 seconds. SAUrg = (0.1, 0.3) ... The urgency of a request to travel between urban and suburban areas is 0.1 or 0.3. ASAVN... Number of SAVs for use throughout the region. SSAVN ... Number of SAVs for suburban parking spots. CSAVN ... Number of SAVs for urban use. SAVN ... Total number of SAVs. Cancel Rate... The cancellation rate predicted by simulation.

[0073] With the above settings, there are two possible SAIntervals and two SAUrgs, meaning there are 2 x 2 = 4 possible scenarios for users traveling between urban and suburban areas. Therefore, the average of the four simulation results is shown as a performance indicator.

[0074] Taking the average of four different scenarios can be interpreted as simulating and predicting performance metrics based on the assumption that four different types of users are present in equal proportions.

[0075] In the example shown in this figure, the lighter the color of the cancellation rate, the lower the cancellation rate and the better the performance. It can be seen that the optimal allocation of units is (ASAVN, SSAVN, CSAVN) = (4, 0, 1).

[0076] Figure 4 is an explanatory diagram showing a series of triangular graphs illustrating the predicted performance indicators for combinations of SAPos=10, SAVN=5, and CAInterval=60, with variations in SAInterval and SAUrg. Figure 5 is an explanatory diagram showing a series of triangular graphs illustrating the predicted performance indicators for combinations of SAPos=20, SAVN=5, and CAInterval=60, with variations in SAInterval and SAUrg. Figure 6 is an explanatory diagram showing a series of triangular graphs illustrating the predicted performance indicators for combinations of SAInterval and SAUrg, with SAPos=10, SAVN=6, and CAInterval=60. Figure 7 is an explanatory diagram showing a series of triangular graphs illustrating the predicted performance indicators for combinations of SAInterval and SAUrg, with SAPos=20, SAVN=6, and CAInterval=60. Figure 8 is an explanatory diagram showing a series of triangular graphs illustrating the predicted performance indicators for combinations of SAInterval and SAUrg, with SAPos=10, SAVN=6, and CAInterval=120. Each example shows a triangular graph illustrating the change in cancellation rates when SAInterval and SAUrg are varied, after the values ​​of SAPos, SAVN, and CAInterval have been set.

[0077] We performed simulations for four possible SAInterval values ​​(90,120), (150,180), (210,240), and (270,300) (from top to bottom in the figure), and for three possible SAUrg values ​​(0.1,0.3), (0.5,0.7), and (0.9,1.1) (from left to right in the figure), to predict the performance indicators.

[0078] Observing these figures, The further suburban spot locations are from urban areas, the more desirable it is to reduce ASAV and increase CSAV. When the number of SAVNs is changed by a small amount (just one), the performance metrics also change by a small amount (similar triangular graphs are obtained). This suggests that it may be possible to interpolate performance metrics by downsampling simulations, or to narrow down the search space when searching for the optimal number of SAVNs. The significant change in the triangular graph when CAInterval is changed from 60 to 120 indicates that the allocation of vehicles needs to be changed between rush hour and off-peak hours.

[0079] (Simulation thinning) Generally, simulations like the one described above often require a considerable amount of computation time. Therefore, when there are many combinations of user characteristics, traffic characteristics, and vehicle allocation, it may be difficult to simulate all of them. In addition, it may be desirable to observe the general trends when the vehicle allocation is changed, narrow down the vehicle allocation options, and then perform a detailed simulation.

[0080] In such cases, For some combinations of user characteristics, traffic characteristics, and vehicle allocation, performance indicators are predicted through simulation. The predicted performance metrics for that portion are used as training data to train the predictive model. For the residual combinations, a trained predictive model is used to predict performance metrics. Therefore, it is also possible to thin out the simulations and perform simulations only for a portion of the unit allocation, while predicting performance indicators for the remaining unit allocations.

[0081] (summary) As described above, the prediction device according to this embodiment is a prediction device that simulates a dispatch system in which multiple geographical constraints that must be met by the pickup point and drop-off point when multiple vehicles pick up and transport users and then drop them off are prepared in advance, and multiple vehicle allocations are prepared in advance in which one of the constraints from the multiple constraints is set for each of the multiple vehicles, For each of the above-mentioned allocations of the number of units, Multiple users who intend to use the ride-hailing system for a journey from a departure point to an arrival point have characteristics related to their selection of whether or not to use the system, depending on the difference between the departure point and the pickup point in the journey, the difference between the arrival point and the drop-off point in the journey, the estimated waiting time from the start of the journey to pickup, and the estimated duration of the journey. Traffic characteristics relating to the traffic in the area to which the aforementioned multiple vehicles and the dispatch system are applied, A prediction unit predicts the performance indicators of the dispatch system for each vehicle allocation by performing the simulation based at least on the above, An illustrative section illustrating the proportion of the multiple constraints in each of the aforementioned allocations and the predicted performance indicators for each of the aforementioned allocations. Configure to include the following:

[0082] Furthermore, in the prediction device according to this embodiment, In the above simulation, We accept requests from simulated users specifying at least the departure and arrival points. From among multiple vehicles, identify a vehicle capable of transporting the simulated user from a pickup point, which is a predetermined vicinity of the departure point, to a drop-off point, which is a predetermined vicinity of the arrival point. The estimated waiting time from when the dispatch request is received until the vehicle arrives at the pickup location, and the estimated time from when the dispatch request is received until the vehicle arrives at the drop-off location or the simulated user arrives at the arrival location, are estimated based on the traffic characteristics. The simulated user is presented with a response that includes at least the pickup and drop-off locations to which the identified vehicle can transport the simulated user, the expected waiting time, and the expected total travel time. Based on the user characteristics related to the selection of the simulated user for whom the response was presented, a decision is made to either dispatch the vehicle related to the response or cancel the dispatch. It can be configured in this way.

[0083] Furthermore, in the prediction device according to this embodiment, The prediction unit predicts the cancellation rate, which is the rate at which dispatches are canceled, as the performance indicator. It can be configured in this way.

[0084] Furthermore, in the prediction device according to this embodiment, The aforementioned multiple vehicles include passenger vehicles capable of transporting multiple users traveling on different itineraries. The prediction unit uses the following performance indicators: The estimated waiting time from the start of the journey to pickup was presented to the simulated user for whom the aforementioned shared vehicle had been assigned. A statistical measure based on the distribution of is predicted as the performance indicator. It can be configured in this way.

[0085] Furthermore, in the prediction device according to this embodiment, The aforementioned multiple vehicles include passenger vehicles capable of transporting multiple users traveling on different itineraries. The prediction unit uses the following performance indicators: The estimated waiting time from the start of the journey to pickup, as presented to the simulated user for whom the aforementioned shared vehicle was dispatched, The simulated waiting time for the simulated user in the simulation A statistical measure based on the distribution of excess time is predicted as the performance indicator. It can be configured in this way.

[0086] Furthermore, in the prediction device according to this embodiment, The aforementioned multiple vehicles include passenger vehicles capable of transporting multiple users traveling on different itineraries. The prediction unit uses the following performance indicators: The estimated travel time presented to the simulated user for whom the aforementioned shared vehicle was dispatched was... The simulated time required for the travel related to the itinerary by the aforementioned simulated user in the simulation. A statistical measure based on the distribution of excess time is predicted as the performance indicator. It can be configured in this way.

[0087] Furthermore, in the prediction device according to this embodiment, The aforementioned allocation of the number of units consists of three types: The predicted cancellation rate relative to the aforementioned proportion is illustrated by the colors plotted on the triangular graph. It can be configured in this way.

[0088] Furthermore, in the prediction device according to this embodiment, The performance indicators are predicted by the simulation for some of the user characteristics, traffic characteristics, and combinations of the multiple vehicle allocations. The predictive model is trained using the performance metrics predicted for the aforementioned portion as training data. For the remaining combinations, the performance indicators are predicted by the trained predictive model. It can be configured in this way.

[0089] Furthermore, the prediction method according to this embodiment is a prediction method in which a prediction device simulates a dispatch system in which multiple geographical constraints that must be met by the pickup point and drop-off point when multiple vehicles pick up and transport users and then drop them off are prepared in advance, and multiple vehicle allocations are prepared in advance in which one of the constraints from the multiple constraints is set for each of the multiple vehicles, For each of the above-mentioned allocations of the number of units, Multiple users who intend to use the ride-hailing system for a journey from a departure point to an arrival point have characteristics related to their selection of whether or not to use the system, depending on the difference between the departure point and the pickup point in the journey, the difference between the arrival point and the drop-off point in the journey, the estimated waiting time from the start of the journey to pickup, and the estimated duration of the journey. Traffic characteristics relating to the traffic in the area to which the aforementioned multiple vehicles and the dispatch system are applied, A prediction step of predicting the performance indicators of the dispatch system for each vehicle allocation by performing the simulation based at least on the above, An illustrating process that illustrates the proportion of the multiple constraints in each of the aforementioned allocations and the predicted performance indicators for each of the aforementioned allocations. Configure to include the following:

[0090] Furthermore, the program according to this embodiment is a program that causes a computer to simulate a dispatch system in which multiple geographical constraints that must be met by the pickup point and drop-off point when multiple vehicles pick up and transport users and then drop them off are provided in advance, and multiple vehicle allocations are provided in advance in which one of the constraints from the multiple constraints is set for each of the multiple vehicles. For each of the above-mentioned allocations of the number of units, Multiple users who intend to use the ride-hailing system for a journey from a departure point to an arrival point have characteristics related to their selection of whether or not to use the system, depending on the difference between the departure point and the pickup point in the journey, the difference between the arrival point and the drop-off point in the journey, the estimated waiting time from the start of the journey to pickup, and the estimated duration of the journey. Traffic characteristics relating to the traffic in the area to which the aforementioned multiple vehicles and the dispatch system are applied, By performing the aforementioned simulation based at least on the above, the performance indicators of the dispatch system for each vehicle allocation can be predicted. The proportion of the multiple constraints in each of the aforementioned allocations and the predicted performance indicators for each of those allocations are shown in the diagram. Configure it as follows.

[0091] This invention allows for various embodiments and modifications without departing from the broad spirit and scope of the invention. Furthermore, the embodiments described above are for illustrative purposes only and do not limit the scope of the invention. In other words, the scope of the invention is indicated not by the embodiments, but by the claims. Various modifications made within the scope of the claims and the equivalent scope of the meaning of the invention are considered to be within the scope of this invention. [Industrial applicability]

[0092] According to the present invention, a prediction device, prediction method, and program can be provided that predict the performance indicators of the vehicle allocation system for a given vehicle allocation by simulating the vehicle allocation system for each vehicle allocation where constraints that the pickup point and drop-off point must satisfy when a user is picked up, transported, and then dropped off are set for multiple vehicles. [Explanation of symbols]

[0093] 101 Prediction device 102 Prediction Section 103 Illustration

Claims

1. A prediction device for simulating a dispatch system in which multiple vehicles are provided with a set of geographical constraints that must be met by the pickup and drop-off points when multiple vehicles pick up and transport users and then drop them off, and multiple vehicle allocations are provided in advance, each of which sets one of the constraints from the set of constraints for the multiple vehicles, For each of the above-mentioned allocations of the number of units, Multiple users who intend to use the ride-hailing system for a journey from a departure point to an arrival point have characteristics related to their selection of whether or not to use the system, depending on the difference between the departure point and the pickup point in the journey, the difference between the arrival point and the drop-off point in the journey, the estimated waiting time from the start of the journey to pickup, and the estimated duration of the journey. Traffic characteristics relating to the traffic in the area to which the aforementioned multiple vehicles and the dispatch system are applied, A prediction unit predicts the performance indicators of the dispatch system for each vehicle allocation by performing the simulation based at least on the above, An illustrative section illustrating the proportion of the multiple constraints in each of the aforementioned allocations and the predicted performance indicators for each of the aforementioned allocations. A prediction device characterized by comprising the following features.

2. A prediction device according to claim 1, wherein in the simulation, We accept requests from simulated users specifying at least the departure and arrival points. From among multiple vehicles, identify a vehicle capable of transporting the simulated user from a pickup point, which is a predetermined vicinity of the departure point, to a drop-off point, which is a predetermined vicinity of the arrival point. The estimated waiting time from when the dispatch request is received until the vehicle arrives at the pickup location, and the estimated time from when the dispatch request is received until the vehicle arrives at the drop-off location or the simulated user arrives at the arrival location, are estimated based on the traffic characteristics. The simulated user is presented with a response that includes at least the pickup and drop-off locations to which the identified vehicle can transport the simulated user, the expected waiting time, and the expected total travel time. Based on the user characteristics related to the selection of the simulated user for whom the response was presented, a decision is made to either dispatch the vehicle related to the response or cancel the dispatch. A prediction device characterized by the following features.

3. A prediction device according to claim 2, The prediction unit predicts the cancellation rate, which is the rate at which dispatches are canceled, as the performance indicator. A prediction device characterized by the following features.

4. A prediction device according to claim 3, The aforementioned multiple vehicles include passenger vehicles capable of transporting multiple users traveling on different itineraries. The prediction unit uses the following performance indicators: The estimated waiting time from the start of the journey to pickup was presented to the simulated user for whom the aforementioned shared vehicle had been assigned. A statistical measure based on the distribution of is predicted as the performance indicator. A prediction device characterized by the following features.

5. A prediction device according to claim 3, The aforementioned multiple vehicles include passenger vehicles capable of transporting multiple users traveling on different itineraries. The prediction unit uses the following performance indicators: The estimated waiting time from the start of the journey to pickup, as presented to the simulated user for whom the aforementioned shared vehicle was dispatched, The simulated waiting time for the simulated user in the simulation A statistical measure based on the distribution of excess time is predicted as the performance indicator. A prediction device characterized by the following features.

6. A prediction device according to claim 3, The aforementioned multiple vehicles include passenger vehicles capable of transporting multiple users traveling on different itineraries. The prediction unit uses the following performance indicators: The estimated travel time presented to the simulated user for whom the aforementioned shared vehicle was dispatched was... The simulated time required for the travel related to the itinerary by the aforementioned simulated user in the simulation. A statistical measure based on the distribution of excess time is predicted as the performance indicator. A prediction device characterized by the following features.

7. The aforementioned allocation of the number of units consists of three types: The predicted cancellation rate relative to the aforementioned proportion is illustrated by the colors plotted on the triangular graph. The prediction device according to any one of claims 4 to 6.

8. The performance indicators are predicted by the simulation for some of the user characteristics, traffic characteristics, and combinations of the multiple vehicle allocations. The predictive model is trained using the performance metrics predicted for the aforementioned portion as training data. For the remaining combinations, the performance indicators are predicted by the trained predictive model. The prediction device according to any one of claims 1 to 6.

9. A prediction method in which a prediction device simulates a dispatch system in which multiple vehicles are provided with a set of geographical constraints that must be met by the pickup and drop-off points when multiple vehicles pick up and transport users and then drop them off, and multiple vehicle allocations are provided in advance, each of which sets one of the constraints from the set of constraints for the multiple vehicles, For each of the above-mentioned allocations of the number of units, Multiple users who intend to use the ride-hailing system for a journey from a departure point to an arrival point have characteristics related to their selection of whether or not to use the system, depending on the difference between the departure point and the pickup point in the journey, the difference between the arrival point and the drop-off point in the journey, the estimated waiting time from the start of the journey to pickup, and the estimated duration of the journey. Traffic characteristics relating to the traffic in the area to which the aforementioned multiple vehicles and the dispatch system are applied, A prediction step of predicting the performance indicators of the dispatch system for each vehicle allocation by performing the simulation based at least on the above, An illustrating process that illustrates the proportion of the multiple constraints in each of the aforementioned allocations and the predicted performance indicators for each of the aforementioned allocations. A prediction method characterized by comprising the following features.

10. A program that causes a computer to simulate a dispatch system in which multiple vehicles are provided with a set of geographical constraints that must be met by the pickup and drop-off points when multiple vehicles pick up and transport passengers and then drop them off, and multiple vehicle allocations are provided in advance, each of which sets one of the constraints from the set of constraints for the multiple vehicles, For each of the above-mentioned allocations of the number of units, Multiple users who intend to use the ride-hailing system for a journey from a departure point to an arrival point have characteristics related to their selection of whether or not to use the system, depending on the difference between the departure point and the pickup point in the journey, the difference between the arrival point and the drop-off point in the journey, the estimated waiting time from the start of the journey to pickup, and the estimated duration of the journey. Traffic characteristics relating to the traffic in the area to which the aforementioned multiple vehicles and the dispatch system are applied, By performing the aforementioned simulation based at least on the above, the performance indicators of the dispatch system for each vehicle allocation can be predicted. The proportion of the multiple constraints in each of the aforementioned allocations and the predicted performance indicators for each of those allocations are shown in the diagram. A program characterized by the following features.

Images