Information processing device, information processing method, and program

Abstract

To provide a delivery route near an optimal solution (an optimal delivery route) of an actual MDVRP.SOLUTION: An information processing device includes a generation unit for generating a plurality of first delivery route candidates for allowing a movable body to start a delivery base and to return to the delivery base for each delivery base, a selection unit for selecting second delivery route candidates having many delivery destinations near the first delivery route candidates as many as movable bodies departing from the delivery base, from among the plurality of first delivery route candidates set to the delivery base, for each delivery base, a classification unit for classifying the delivery destinations on the basis of distance between the delivery destinations and the second delivery route candidates for each selected second delivery route candidate and generating a delivery destination group for each movable body, and a determination unit for determining a delivery route for patrolling all delivery destinations in the delivery destination group without omission for each delivery destination group.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present disclosure relates to an information processing device, an information processing method, and a program for generating a delivery route. [Background technology]

[0002] A method of finding an approximate solution using a clustering method is known as a solution method for the Multi-Depots Vehicle Routing Problem (MDVRP), which is a problem of planning routes for delivering goods from multiple delivery depots to multiple delivery destinations using multiple vehicles.

[0003] As a related technique, Patent Document 1 discloses a route derivation system that utilizes a quantum computer to derive a desirable travel route for each moving body when the moving bodies split up and visit multiple destinations. The classical computer unit of the route derivation system of Patent Document 1 derives a tentative travel route for each moving body when the moving bodies split up and visit multiple destinations. The classical computer unit also causes an Ising machine to execute a solution-finding process for optimizing each of the tentative travel routes for the moving bodies so that the travel cost of each moving body is minimized. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent Publication No. 2022-182286 Summary of the Invention [Problem to be solved by the invention]

[0005] However, conventional clustering methods consider only some of the factors that are to be determined by MDVRP, such as (1) which vehicle should depart and arrive from which distribution center, (2) which vehicle should deliver to which destination, and (3) what order each destination should be delivered to.

[0006] Therefore, the solution (delivery route) obtained using the conventional clustering method is significantly different from the optimal solution (optimal delivery route) of the actual MDVRP.

[0007] One example of the objective of the present disclosure is to provide a delivery route that is close to an optimal solution (optimum delivery route) of the actual MDVRP. [Means for solving the problem]

[0008] In order to achieve the above object, an information processing device according to one aspect of the present disclosure includes: A generation unit that generates, for each delivery base, a plurality of first delivery route candidates along which a mobile object departs from the delivery base and returns to the delivery base; A selection unit that selects, for each of the delivery bases, from among the plurality of first delivery route candidates set at the delivery base, a second delivery route candidate having many delivery destinations near the first delivery route candidate in a number equal to the number of mobile objects departing and arriving at the delivery base; A classification unit that classifies the delivery destinations for each of the selected second delivery route candidates based on a proximity between the second delivery route candidate and the delivery destinations, and generates a delivery destination group for each of the moving objects; A determination unit that determines, for each of the delivery destination groups, a delivery route that travels to all of the delivery destinations in the delivery destination group without omission; The present invention is characterized by having the following.

[0009] In order to achieve the above object, an information processing method according to one aspect of the present disclosure includes: Generate, for each delivery base, a plurality of first delivery route candidates along which a mobile object departs from the delivery base and returns to the delivery base; For each of the delivery bases, from among the plurality of first delivery route candidates set at the delivery base, a second delivery route candidate having many delivery destinations near the first delivery route candidate is selected in a number corresponding to the number of mobile objects departing from and arriving at the delivery base; For each of the selected second delivery route candidates, classify the delivery destinations based on the proximity of the second delivery route candidate to the delivery destinations, and generate a delivery destination group for each of the moving objects; determining, for each of the delivery destination groups, a delivery route that visits all of the delivery destinations in the delivery destination group without omission; It is characterized by:

[0010] Furthermore, in order to achieve the above object, a program according to one aspect of the present disclosure includes: generating, for each delivery base, a plurality of first delivery route candidates along which a mobile object departs from the delivery base and returns to the delivery base; For each of the delivery bases, from among the plurality of first delivery route candidates set at the delivery base, a second delivery route candidate having many delivery destinations near the first delivery route candidate is selected in a number corresponding to the number of mobile objects departing from and arriving at the delivery base; For each of the selected second delivery route candidates, the delivery destinations are classified based on the proximity of the second delivery route candidate to the delivery destinations, and a delivery destination group is generated for each of the moving objects; determining, for each of the delivery destination groups, a delivery route that travels to all of the delivery destinations in the delivery destination group without omission; It is characterized by: Effect of the Invention

[0011] As described above, according to the present disclosure, it is possible to provide a delivery route that is close to the optimal solution (optimum delivery route) of the actual MDVRP. [Brief description of the drawings]

[0012] [Figure 1] FIG. 1 is a diagram for explaining an example of optimization of a delivery route. [Diagram 2] FIG. 2 is a diagram illustrating an example of an information processing device. [Diagram 3] FIG. 3 is a diagram illustrating an example of a system having an information processing device. [Figure 4] FIG. 4 is a diagram for explaining an example of the first delivery route candidate. [Diagram 5] FIG. 5 is a diagram for explaining an example of a first delivery route candidate when there are a plurality of delivery bases. [Figure 6] FIG. 6 is a diagram for explaining an example of a score when one second delivery route candidate is selected. [Figure 7] FIG. 7 is a diagram for explaining an example of the scores and the return values ​​of the objective function when two second delivery route candidates are selected. [Figure 8] FIG. 8 is a diagram for explaining an example of generating a destination group. [Figure 9] FIG. 9 is a diagram for explaining an example of decision variables and calculation results. [Figure 10] FIG. 10 is a diagram illustrating an example of the operation of the information processing device. [Figure 11] FIG. 11 is a diagram for explaining an example of the scores and the return values ​​of the objective function when two second delivery route candidates are selected. [Figure 12] FIG. 12 is a diagram illustrating an example of a computer that realizes the information processing device in the embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] First, an overview will be given to facilitate understanding of the embodiments described below. FIG. 1 is a diagram for explaining an example of optimizing a delivery route. In the example of FIG. 1, the delivery route is optimized using four steps. Also, the star marks (★) in FIG. 1 represent delivery bases (depots) for delivery, and the black circles (●) represent delivery destinations. Furthermore, the dashed lines represent candidate delivery routes departing from and arriving at the delivery bases. Note that in the example of FIG. 1, it is assumed that there are three moving objects. The moving objects are, for example, vehicles, aircraft, ships, etc.

[0014] In step 1 (generation process), as shown in A of FIG. 1, a rough route (first delivery route candidate: dashed line) of a mobile object departing from a delivery base and returning to the delivery base is generated.

[0015] In step 2 (selection process), as shown in B of Fig. 1, a second delivery route candidate corresponding to each delivery base (or mobile object) is selected from the first delivery route candidates. Specifically, a second delivery route candidate (thick solid lines (1)(2)(3)) is selected for each of the three mobile objects.

[0016] In step 3 (classification process), the delivery destinations are classified based on the second delivery route candidates, as shown in C of Fig. 1. Specifically, the delivery destinations are assigned to the thick solid lines (1), (2), and (3) that represent the second delivery route candidates, and the delivery destinations are classified (grouped) to generate delivery destination groups (clusters: solid line ranges).

[0017] In step 4 (decision processing), as shown in D of Fig. 1, a delivery route is determined for each destination group (cluster) so that the mobile object can visit all destinations in the destination group without missing any. Specifically, a traveling salesman problem is solved for each destination group to determine the delivery route.

[0018] By carrying out the above-mentioned steps 1 to 4, it is possible to provide a delivery route that is close to the optimal solution (optimum delivery route) of the actual MDVRP.

[0019] Next, an embodiment will be described with reference to the drawings. In the drawings described below, elements having the same or corresponding functions are denoted by the same reference numerals, and repeated description thereof may be omitted.

[0020] (Embodiment) The configuration of an information processing device (route generating device) that generates a delivery route in an embodiment will be described with reference to Fig. 2. Fig. 2 is a diagram for explaining an example of the information processing device.

[0021] [Device configuration] The information processing device shown in Fig. 2 is a device that provides a delivery route close to the optimal solution (optimum delivery route) of the actual MDVRP. As shown in Fig. 2, the information processing device 10 includes a generation unit 11, a selection unit 12, a classification unit 13, and a determination unit 14.

[0022] The generation unit 11 generates, for each delivery base, a plurality of first delivery route candidates along which a mobile object departs from the delivery base and returns to the delivery base.

[0023] The selection unit 12 selects, for each delivery base, from among a plurality of first delivery route candidates set at the delivery base, second delivery route candidates having many delivery destinations near the first delivery route candidates in the same number as the number of mobile bodies departing from and arriving at the delivery base.

[0024] The classification unit 13 classifies the delivery destinations for each selected second delivery route candidate based on the proximity between the second delivery route candidate and the delivery destinations, and generates a delivery destination group for each moving object.

[0025] The determination unit 14 determines, for each delivery destination group, a delivery route that travels to all delivery destinations in the delivery destination group without omission.

[0026] In this manner, in the embodiment, a second delivery route candidate is selected from a plurality of first delivery route candidates set for each delivery base, delivery destinations are classified using the second delivery route candidate, and a delivery route is determined for each classified delivery destination group, so that factors to be determined by MDVRP can be taken into consideration more than in the past. As a result, a delivery route closer to the actual optimal solution (optimal delivery route) of MDVRP can be provided.

[0027] [System Configuration] Next, the configuration of the information processing device 10 in the embodiment will be described in more detail with reference to Fig. 3. Fig. 3 is a diagram for explaining an example of a system having an information processing device.

[0028] 3 includes an information processing device 10, a quantum computer 20, a terminal device 30, and an output device 40. The information processing device 10, the quantum computer 20, the terminal device 30, and the output device 40 are connected to each other via a network.

[0029] The network is a general network constructed using communication lines such as the Internet, a LAN (Local Area Network), a dedicated line, a telephone line, an in-house network, a mobile communication network, Bluetooth (registered trademark), or Wi-Fi (Wireless Fidelity) (registered trademark).

[0030] The information processing device 10 is, for example, an information processing device such as a CPU (Central Processing Unit), a programmable device such as an FPGA (Field-Programmable Gate Array), a GPU (Graphics Processing Unit), or a circuit equipped with any one or more of these, a server computer, or the like.

[0031] The quantum computer 20 is hardware for solving combinatorial optimization problems. Combinatorial optimization problems are solved using, for example, a QUBO (Quadratic Unconstrained Binary Optimization) solver. QUBO solvers include, for example, quantum annealing and pseudo-quantum annealing.

[0032] The terminal device 30 is, for example, an information processing device such as a CPU, a programmable device such as an FPGA, a GPU, or a circuit equipped with any one or more of them, a personal computer, or a mobile terminal.

[0033] The output device 40 acquires output information, which will be described later, that has been converted into an outputtable format, and outputs images and sounds generated based on the output information. The output device 40 is, for example, an image display device using a liquid crystal, an organic EL (Electro Luminescence), or a CRT (Cathode Ray Tube). Furthermore, the image display device may also include an audio output device such as a speaker. The output device 40 may also be a printing device such as a printer.

[0034] The information processing device 10 will be described in detail. The information processing device 10 includes a generating unit 11, a selecting unit 12, a classifying unit 13, a determining unit 14, and an output information generating unit 15.

[0035] The generating unit 11 (generation process) first acquires delivery information stored in a storage device (not shown). The delivery information includes at least delivery base location coordinate information indicating the location coordinates of the delivery base, delivery destination location coordinate information indicating the location coordinates of the delivery destination, and mobile object information indicating the number of mobile objects. The delivery information may also include delivery base-destination distance information indicating the distance between the delivery base and the delivery destination, and delivery destination distance information indicating the distance between the delivery destinations.

[0036] The distance between the delivery base and the delivery destination may be calculated using the location coordinates of the delivery base and the location coordinates of the delivery destination. The distance between delivery destinations may be calculated using the location coordinates of the delivery destination. The delivery base / mobile object related information includes, for example, information indicating the number of mobile objects departing from and arriving at the delivery base. The delivery information may also include delivery base / mobile object related information indicating the association between the delivery base and the mobile object.

[0037] Next, the generation unit 11 uses the delivery base location coordinate information and the delivery destination location coordinate information to generate, for each delivery base, a plurality of first delivery route candidates in which a predetermined number of mobile bodies depart from the delivery base and return to the delivery base.

[0038] Example A in Figure 1 shows multiple straight-line routes (dashed lines) generated radially from a distribution center (★). A straight-line route is a route in which a mobile object leaves a distribution center (★), arrives at a designated location, and then returns to the distribution center (★) via the same route.

[0039] Moreover, the first delivery route candidate is not limited to the above-mentioned straight-line route. The shape of the first delivery route candidate may be, for example, a triangle, a circle, an ellipse, or the like. FIG. 4 is a diagram for explaining an example of a first delivery route candidate. A in FIG. 4 shows a first delivery route candidate that is a triangular route. B in FIG. 4 shows a first delivery route candidate that is a circular route. It is preferable that the route lengths of the first delivery route candidates are approximately the same. Although the example in FIG. 4 shows an example using a simple figure, a route that takes into account actual roads, topography, and current conditions may also be used.

[0040] Furthermore, as shown in FIG. 5, a plurality of first delivery route candidates may be set for each of a plurality of delivery bases (★). FIG. 5 is a diagram for explaining an example of first delivery route candidates when there are a plurality of delivery bases. In the example of FIG. 5, a plurality of first delivery route candidates c1 to c10 are set for each of the two delivery bases (★). In the example of FIG. 5, a decision variable xc (x1 to x 10 , x 11 x 20 : dashed line).

[0041] The selection unit 12 (selection process) first detects the number of mobile objects departing from and arriving at a delivery center using the delivery center-mobile object related information. Next, for each delivery center, the selection unit 12 selects a second delivery route candidate having many delivery destinations near the first delivery route candidate from among a plurality of first delivery route candidates for the delivery center.

[0042] The selection process uses, for example, a QUBO solver to select a second delivery route candidate. The QUBO solver uses quantum annealing, pseudo quantum annealing, etc. The Hamiltonian H1 input to the QUBO solver is shown in Equation 1.

[0043]

number

[0044] Constructing the Hamiltonian H1 shown in Equation 1, Constr1 represents a constraint condition, and obj1′ represents an objective function. Note that α represents a weighting parameter (positive constant). α is determined, for example, by experiments, simulations, etc.

[0045] The constraint condition Constr1 represents a condition for selecting second delivery route candidates, the number of which is equal to the number of mobile units, from among a plurality of first delivery route candidates. The constraint condition Constr1 can be expressed as in Equation 2.

[0046]

number

[0047] x c is a decision variable that represents each of the first delivery route candidates. c is a positive integer that represents the number of the first delivery route candidate. c is a binary variable that is "1" if the first delivery route candidate is selected and "0" if it is not selected. C R represents multiple first delivery route candidates (a set of first delivery route candidates). V represents the number of mobile objects.

[0048] The objective function obj1' is a function that adopts the first delivery route candidate that is close to as many delivery destinations as possible. The objective function obj1' can also be said to be a function that increases the average score of as many delivery destinations as possible. The objective function obj1' is a proxy function that approximates the objective function obj1 shown in Equation 3 using a machine learning technique called FM (Factorization Machine).

[0049]

number

[0050] The reason for using a proxy function is that the objective function obj1 is c C R This is because it is a high-order formula that includes power terms and is generally difficult to optimize.

[0051] D in the objective function obj1 represents a set of delivery destinations d. In addition, the score function S(c, d) of the objective function obj1 is a function that returns a score to be assigned to each delivery destination d for each selected first delivery route candidate c.

[0052] The score is a function that satisfies 0≦S(c,d)≦1 for the first delivery route candidate c and the delivery destination d. In addition, the closer the distance between the first delivery route candidate c and the delivery destination d is, the closer the score is to 1.

[0053] Fig. 6 is a diagram for explaining an example of a score when selecting one second delivery route candidate. In Fig. 6A, one delivery base (★), three first delivery route candidates (straight-line routes) c1, c2, c3, and five delivery destinations d1, d2, d3, d4, d5 (●) are shown. Then, by calculating the scores of the delivery destinations d1 to d5 for each of the first delivery route candidates c1, c2, c3, the result shown in Fig. 6B can be obtained.

[0054] FIG. 6B shows the scores of the delivery destinations for each of the first delivery route candidates c1, c2, and c3. In the case of the first delivery route candidate c1, the scores of the delivery destinations d2 to d5 are higher than a preset score, so it can be determined that the delivery destinations d2 to d5 are near the first delivery route candidate c1. In the case of the first delivery route candidate c2, the scores of the delivery destinations d1 and d5 are higher than a preset score, so it can be determined that the delivery destinations d1 and d5 are near the first delivery route candidate c2. In the case of the first delivery route candidate c3, the scores of the delivery destinations d1 to d5 are lower than a preset score, so it can be determined that the delivery destinations d1 to d5 are not near the first delivery route candidate c3.

[0055] As described above, the score function S(c, d) is a function that satisfies 0≦S(c, d)≦1, so F1 of the objective function obj1 shown in Equation 4 is also a function that satisfies 0≦F1≦1.

[0056]

number

[0057] Next, in the example of A in Fig. 6, a case where two second delivery route candidates are selected using the objective function obj1 will be described. Fig. 7 is a diagram for explaining an example of the score and the return value of the objective function when selecting two second delivery route candidates. In the example of Fig. 7, the return value (-4.50) of the objective function obj1 corresponding to the first delivery route candidate (c1, c2) is the smallest, so c1 and c2 are selected as the second delivery route candidates.

[0058] Next, we explain how to convert the objective function obj1 to the objective function obj1' using FM. FM is a type of learning algorithm called supervised learning. For more information on FM, please refer to the following references [1] to [3].

[0059] [1] Steffen Rendle, “Factorization Machines”, [online], December 13-17, 2010, IEEE International Conference on Data Mining, [Retrieved June 9, 2023], Internet<URL:https: / / www.csie.ntu.edu.tw / ~b97053 / paper / Rendle2010FM.pdf>

[0060] [2] Koki Kitai, Jiang Guo, Shenghong Ju, Shu Tanaka, Koji Tsuda, Junichiro Shiomi, and Ryo Tamura, “Designing metamaterials with quantum annealing and factorization machines”, [online], March 16, 2020, PHYSICAL REVIEW RESEARCH Volume 2, Issue 1, [Retrieved June 9, 2023], Internet<URL:https: / / journals.aps.org / prresearch / pdf / 10.1103 / PhysRevResearch.2.013319>

[0061] [3] Munemitsu Tanaka, Masashi Yamashita, Yuya Seki, “Black-box optimization using annealing machines”, [online], 2022, Journal of the Japanese Society for Neural Networks, Vol. 29, No. 4, p. 164-173, [Retrieved June 9, 2023], Internet<URL:https: / / www.jstage.jst.go.jp / article / jnns / 29 / 4 / 29_164 / _pdf>

[0062] An objective function obj1′ when selecting a second delivery route candidate from a plurality of first delivery route candidates is shown in Equation 5.

[0063]

number

[0064] In addition, w in number 5 c , v c,k are model parameters determined by learning using FM. K represents the factorization size (hyperparameter). x c1 , x c2 is a decision variable that represents the first delivery route candidate to be selected. The objective function obj1 shown in Equation 3 is c However, in Equation 5, it is approximated as the sum of products of variables representing any two delivery route candidates, making it an equation that can be optimized.

[0065] Hamiltonian H1 to be input to the QUBO solver when selecting a second delivery route candidate from multiple first delivery route candidates is shown in Formula 6.

[0066]

number

[0067] When the selection unit 12 inputs the Hamiltonian H1 to the QUBO solver, it obtains the calculation result of the quantum computer 20 from the QUBO solver. The calculation result includes the decision variable x c The value for each (either "0" or "1") is returned. Then, the selection unit 12 selects the decision variable x c The first delivery route candidate with a value of "1" is set as the second delivery route candidate.

[0068] In the example of Figure 5, the decision variables x2, x 16 , x 17 The value of becomes “1”, and the second delivery route candidate (1), (2), and (3) are selected, as shown in FIG. 1B.

[0069] The classification unit 13 (classification process) calculates the distance between the delivery destination and the second delivery route candidate for each selected second delivery route candidate, associates the second delivery route candidate with the delivery destination having the shortest calculated distance, and generates a delivery destination group for each second delivery route candidate.

[0070] Fig. 8 is a diagram for explaining an example of generating a delivery destination group. In the example of Fig. 8, as shown in A of Fig. 8, for each of the second delivery route candidates (1), (2), and (3), a delivery destination close to the second delivery route candidate is selected, and three delivery destination groups are generated as shown in B of Fig. 8.

[0071] The classification process is not limited to the above-mentioned classification method, and any process that can classify delivery destinations based on the second delivery route candidates may be used. Therefore, a Hamiltonian for the classification process may be constructed, and the classification may be performed using a QUBO solver or the like.

[0072] The determination unit 14 (determination process) determines, for each destination group, a delivery route that visits all destinations in the destination group without missing any, for example, by using the TSP (Traveling Salesman Problem), etc. Specifically, since various methods are known for solving the Traveling Salesman Problem, it is basically sufficient to utilize an existing method.

[0073] Here, a method for solving the traveling salesman problem using a QUBO solver will be described. Specifically, the determination unit 14 inputs the Hamiltonian H2 shown in Equation 7 to the QUBO solver to solve the traveling salesman problem.

[0074]

number

[0075] Constr2 and Constr3 of the Hamiltonian H2 shown in Equation 7 represent constraint conditions, and obj2 represents the objective function. Note that β and γ represent weight parameters (positive constants). β and γ are determined, for example, by experiments or simulations.

[0076] The constraint condition Constr2 shown in Expression 8 indicates that the delivery destination d can be visited only once. The constraint condition Constr2 can be expressed as Expression 8.

[0077]

number

[0078] x i,d is a binary variable that takes the value "1" if the i-th delivery destination d is visited and "0" if it is not visited. c represents the destination group (cluster c). Dc represents the number of delivery destinations d in cluster c.

[0079] The constraint Constr3 shown in Equation 9 indicates that only one delivery destination d can be visited at a time.

[0080]

number

[0081] The objective function obj2 is a function that adopts a route with a short delivery distance. The objective function obj2 can be expressed as in Equation 10.

[0082]

number

[0083] F2 in the objective function obj2 represents the distance from the delivery point P to the first delivery destination. F3 represents the distance between delivery destinations (the sum of the distances between the i-th and i+1-th delivery destinations). F4 represents the distance from the last delivery destination to the delivery point P. d1 and d2 represent any delivery destinations.

[0084] When the determination unit 14 inputs the Hamiltonian H2 to the QUBO solver, it obtains the calculation result of the quantum computer 20 from the QUBO solver. Fig. 9 is a diagram for explaining an example of the decision variables and the calculation result.

[0085] A in FIG. 9 shows an example of a decision variable. In the example in A in FIG. 9, the horizontal axis (i) shows "1", "2", "3", "4", "5", etc., which indicate the order in which the delivery destinations are visited, and the vertical axis (Dst=d) shows "Dst1", "Dst2", "Dst3", "Dst4", "Dst5", etc., which indicate the delivery destinations. The decision variable is x i,d It is shown in the form:

[0086] FIG. 9B shows an example of the calculation result. In the example of FIG. 9B, the decision variable x i,d The order of the destinations where the value of "1" is "Dst2", "Dst1", "Dst4", "Dst5", "Dst3", etc. is shown. i,d The sum of the vertical axis and the decision variable x i,d The sum of these is "1".

[0087] By determining the visiting order as described above, a delivery route that visits all of the destinations without missing any can be determined for each of the three destination groups in B of Figure 8 (C of Figure 1), as shown in D of Figure 1.

[0088] The output information unit 15 generates output information for outputting to the output device 40, for example, a display of a first delivery route candidate shown in A of FIG. 1, a display of a second delivery route candidate shown in B of FIG. 1, a display of delivery destination groups shown in C of FIG. 1, and a display of delivery routes for each delivery destination group shown in D of FIG. 1.

[0089] In addition, the output information unit 15 may generate output information for outputting to the output device 40 the number of delivery bases, the number of mobile objects, the number of delivery destinations, the processing time from step 1 to step 4, the number of delivery destinations for each mobile object, the order in which the delivery destinations are visited, the distance of the delivery route, the time required for delivery, etc.

[0090] [Device operation] Next, the operation of the information processing device in the embodiment will be described with reference to FIG. 10. FIG. 10 is a diagram for explaining an example of the operation of the information processing device. In the following description, the diagram will be referred to as appropriate. Also, in the embodiment, the information processing method is implemented by operating the information processing device. Therefore, the description of the information processing method in the embodiment will be replaced with the following description of the operation of the information processing device.

[0091] As shown in FIG. 10, first, the generating unit 11 (generation process) generates, for each distribution base, a plurality of first delivery route candidates along which a mobile object departs from the distribution base and returns to the distribution base (step A1).

[0092] Specifically, in step A1, the generation unit 11 first acquires delivery information stored in a storage device (not shown). Next, in step A1, the generation unit 11 uses the delivery base location coordinate information and the delivery destination location coordinate information to generate, for each delivery base, a plurality of first delivery route candidates along which a preset number of mobile bodies depart from the delivery base and return to the delivery base.

[0093] Next, the selection unit 12 (selection process) selects, for each distribution base, from among a plurality of first distribution route candidates set at the distribution base, a second distribution route candidate that has many distribution destinations near the first distribution route candidate, the number of which corresponds to the number of mobile objects departing from and arriving at the distribution base (step A2).

[0094] Specifically, in step A2, the selection unit 12 first detects the number of mobile objects departing from and arriving at a delivery center using the delivery center-mobile object related information. Next, in step A2, the selection unit 12 selects, for each delivery center, a second delivery route candidate having many delivery destinations near the first delivery route candidate from among a plurality of first delivery route candidates for the delivery center.

[0095] For example, the selection unit 12 inputs the Hamiltonian H1 to the QUBO solver as described above, obtains the calculation result of the quantum computer 20 from the QUBO solver, and selects the decision variable x c The first delivery route candidate with a value of "1" is set as the second delivery route candidate.

[0096] Next, the classification unit 13 (classification process) classifies the delivery destinations for each selected second delivery route candidate based on the proximity between the second delivery route candidate and the delivery destination, and generates a delivery destination group for each moving object (step A3).

[0097] Specifically, in step A3, the classification unit 13 calculates the distance between the delivery destination and the second delivery route candidate for each selected second delivery route candidate, performs a classification process to associate the delivery destination with the second delivery route candidate having the shortest calculated distance, and generates a delivery destination group for each second delivery route candidate (mobile object).

[0098] Next, the determination unit 14 (determination process) determines, for each delivery destination group, a delivery route that travels to all delivery destinations in the delivery destination group without omission (step A4).

[0099] Specifically, in step A4, the determination unit 14 determines, for each destination group, a delivery route that visits all destinations in the destination group without missing any, for example, by using the TSP (Traveling Salesman Problem), etc. Specifically, since various methods are known for solving the Traveling Salesman Problem, it is basically sufficient to utilize an existing method.

[0100] Next, the output information unit 15 generates output information for outputting to the output device 40, for example, a display of a first delivery route candidate shown in A of FIG. 1, a display of a second delivery route candidate shown in B of FIG. 1, a display of delivery destination groups shown in C of FIG. 1, and a display of delivery routes for each delivery destination group shown in D of FIG. 1 (step A5).

[0101] In addition, in step A5, the output information unit 15 may generate output information for causing the output device 40 to output the number of delivery bases, the number of mobile objects, the number of delivery destinations, the processing time from step 1 to step 4, the number of delivery destinations for each mobile object, the order in which the delivery destinations are visited, the distance of the delivery route, the time required for delivery, etc.

[0102] [Effects of the embodiment] As described above, according to the embodiment, the second delivery route candidate is selected from a plurality of first delivery route candidates set for each delivery base, the second delivery route candidate is used to classify delivery destinations, and a delivery route is determined for each classified delivery destination group, so that factors to be determined by MDVRP can be taken into consideration more than in the past. As a result, a delivery route closer to the actual optimal solution (optimal delivery route) of MDVRP can be provided.

[0103] (Variation 1) In the first modification, when the number of moving objects for each delivery base is predetermined, the constraint Constr1 of the Hamiltonian H1 is replaced by the formula 11. v,p represents the number of mobile units at delivery point p, and c r、p is the first set of candidate delivery routes starting from and ending at delivery point p.

[0104]

number

[0105] (Variation 2) In the second modification, the selection unit 12 (selection process) selects the objective function obj1 as the objective function obj 11 Change to the objective function obj. 11 is the score function S(c,d)x for each of the first delivery route candidates c. c The max function calculates and returns the maximum value among them.

[0106]

number

[0107] FIG. 11 is a diagram for explaining an example of the score and the return value of the objective function when selecting two second delivery route candidates. In the example of FIG. 11, in A of FIG. 6, the objective function obj corresponding to the first delivery route candidate (c1, c2) 11 Since the return value (-4.15) of is the smallest, c1 and c2 are selected as the second delivery route candidates.

[0108] Next, the Hamiltonian includes the objective function obj shown in Equation 12. 11 The surrogate function (objective function) obj is approximated using a machine learning technique called Factorization Machine (FM). 11 ' is used.

[0109] Also in the first modification, the selection unit 12 selects the second delivery route candidate by using a QUBO solver or the like. As the QUBO solver, quantum annealing, pseudo quantum annealing, or the like is used. The Hamiltonian H 11 is shown in Equation 13.

[0110]

number

[0111] Objective function obj when selecting a second delivery route candidate from multiple first delivery route candidates 11 ' is shown in Equation 14.

number

[0112] In addition, w in equation 14 c , v c,k are model parameters determined by learning using FM. K represents the factorization size (hyperparameter). x c1 , x c2 is a decision variable that represents the first candidate delivery route to be selected.

[0113] The selection unit 12 selects the Hamiltonian H 11 When the above equation is input to the QUBO solver, the computation result of the quantum computer 20 is obtained from the QUBO solver. The computation result includes the decision variable x c The value for each (either "0" or "1") is returned. Then, the selection unit 12 selects the decision variable x c The first delivery route candidate with a value of "1" is set as the second delivery route candidate.

[0114] [program] The program in the embodiment may be a program that causes a computer to execute steps A1 to A5 shown in Fig. 10. By installing and executing this program in a computer, the information processing device and information processing method in the embodiment can be realized. In this case, the processor of the computer functions as a generating unit 11, a selecting unit 12, a classifying unit 13, a deciding unit 14, and an output information generating unit 15 to perform processing.

[0115] The program in the embodiment may be executed by a computer system constructed by a plurality of computers. In this case, for example, each computer may function as any one of the generating unit 11, the selecting unit 12, the classifying unit 13, the determining unit 14, and the output information generating unit 15.

[0116] [Physical configuration] Here, a computer that realizes an information processing device by executing a program in the embodiment will be described with reference to Fig. 12. Fig. 12 is a diagram for explaining an example of a computer that realizes an information processing device in the embodiment.

[0117] 12, the computer 110 includes a CPU (Central Processing Unit) 111, a main memory 112, a storage device 113, an input interface 114, a display controller 115, a data reader / writer 116, and a communication interface 117. These components are connected to each other via a bus 121 so as to be able to communicate data with each other. Note that the computer 110 may include a GPU or an FPGA in addition to the CPU 111 or instead of the CPU 111.

[0118] The CPU 111 loads a program in the embodiment, which is composed of a group of codes and is stored in the storage device 113, into the main memory 112, and executes each code in a predetermined order to perform various calculations. The main memory 112 is typically a volatile storage device such as a DRAM (Dynamic Random Access Memory).

[0119] Moreover, the program in the embodiment is provided in a state stored in a computer-readable recording medium 120. Note that the program in the embodiment may be distributed on the Internet connected via the communication interface 117.

[0120] Specific examples of the storage device 113 include a hard disk drive and a semiconductor storage device such as a flash memory. The input interface 114 mediates data transmission between the CPU 111 and input devices 118 such as a keyboard and a mouse. The display controller 115 is connected to a display device 119 and controls the display on the display device 119.

[0121] Data reader / writer 116 mediates data transmission between CPU 111 and recording medium 120, reads programs from recording medium 120, and writes processing results in computer 110 to recording medium 120. Communication interface 117 mediates data transmission between CPU 111 and other computers.

[0122] Specific examples of recording medium 120 include general-purpose semiconductor storage devices such as CF (Compact Flash (registered trademark)) and SD (Secure Digital), magnetic recording media such as a flexible disk, or optical recording media such as a CD-ROM (Compact Disk Read Only Memory).

[0123] The information processing device 10 in the embodiment can be realized not by a computer with a program installed, but by hardware corresponding to each part, for example, an electronic circuit. Furthermore, the information processing device 10 may be partially realized by a program and the remaining part by hardware. In the embodiment, the computer is not limited to the computer shown in FIG. 12.

[0124] Although the invention has been described above with reference to the embodiments, the invention is not limited to the above-described embodiments. Various modifications that can be understood by those skilled in the art can be made to the configuration and details of the invention within the scope of the invention. [Industrial Applicability]

[0125] According to the above description, it is possible to provide a delivery route that is close to the optimal solution (optimum delivery route) of the actual MDVRP. Also, it is useful in fields where MDVRP is required. [Explanation of symbols]

[0126] 10. Information processing device 11 Generation part 12 Selection section 13 Classification section 14 Decision Section 15 Output information generation unit 20. Quantum Computer 30 Terminal Equipment 40 Output Device 100 Systems 110 Computer 111 CPU 112 Main memory 113 Storage device 114 Input Interface 115 Display Controller 116 Data Reader / Writer 117 Communication Interface 118 Input Devices 119 Display device 120 Recording media 121 Bus

Claims

1. A generation means for generating, for each delivery base, a plurality of first delivery route candidates along which a mobile object departs from the delivery base and returns to the delivery base; A selection means for selecting, for each of the delivery bases, from among the plurality of first delivery route candidates set at the delivery base, a second delivery route candidate having many delivery destinations near the first delivery route candidate in a number equal to the number of mobile objects departing and arriving at the delivery base; A classification means for classifying the delivery destinations for each of the selected second delivery route candidates based on the proximity of the delivery destinations to the second delivery route candidates, and generating a delivery destination group for each of the moving objects; A determination means for determining, for each of the delivery destination groups, a delivery route that travels around all of the delivery destinations in the delivery destination group without omission; An information processing device having the above configuration.

2. The selection means selects the second delivery route candidate using a quadratic unconstrained binary optimization (QUBO) solver. The information processing device according to claim 1 .

3. The Hamiltonian input to the QUBO solver is composed of a constraint condition for selecting the second delivery route candidates from the plurality of first delivery route candidates, the number of which is equal to the number of the moving objects, and an objective function for adopting the first delivery route candidates that are close to as many of the delivery destinations as possible. The information processing device according to claim 2 .

4. The constraint Constr １ is expressed by Equation 1, [0010] x ｃ : Decision variables representing each of the first delivery route candidates c: the number of the first delivery route candidate C Ｒ : A plurality of the first delivery route candidates N Ｖ : The number of the moving objects The objective function obj1′ is expressed by Equation 2: [0025] x ｃ１ , x ｃ２ : A decision variable representing the first delivery route candidate to be selected W ｃ , v ｃ，ｋ : Model parameters determined by learning the Factorization Machine K: Factorization size (hyperparameter) The information processing device according to claim 3 .

5. The first delivery route candidate is a straight-line route in which the mobile object departs from the delivery base, and when the mobile object arrives at a predetermined position, returns to the delivery base via the same route. The information processing device according to claim 1 .

6. The classification means calculates a distance between the delivery destination and the second delivery route candidate for each of the selected second delivery route candidates, associates the second delivery route candidate with the delivery destination having the shortest calculated distance, and generates the delivery destination group for each of the second delivery route candidates. The information processing device according to claim 1 .

7. the determining means determines, for each of the delivery destination groups, the delivery route that visits all of the delivery destinations in the delivery destination group without omission, by using a traveling salesman problem; The information processing device according to claim 1 .

8. The traveling salesman problem is solved using the QUBO solver. The information processing device according to claim 7.

9. An information processing device, Generate, for each delivery base, a plurality of first delivery route candidates along which a mobile object departs from the delivery base and returns to the delivery base; For each of the delivery bases, from among the plurality of first delivery route candidates set at the delivery base, a second delivery route candidate having many delivery destinations near the first delivery route candidate is selected in a number corresponding to the number of mobile objects departing from and arriving at the delivery base; For each of the selected second delivery route candidates, classify the delivery destinations based on the proximity of the second delivery route candidate to the delivery destinations, and generate a delivery destination group for each of the moving objects; determining, for each of the delivery destination groups, a delivery route that visits all of the delivery destinations in the delivery destination group without omission; Information processing methods.

10. On the computer, generating, for each delivery base, a plurality of first delivery route candidates along which a mobile object departs from the delivery base and returns to the delivery base; For each of the delivery bases, from among the plurality of first delivery route candidates set at the delivery base, a second delivery route candidate having many delivery destinations near the first delivery route candidate is selected in a number corresponding to the number of mobile objects departing and arriving at the delivery base; For each of the selected second delivery route candidates, the delivery destinations are classified based on the proximity of the second delivery route candidate to the delivery destinations, and a delivery destination group is generated for each of the mobile objects. determining, for each of the delivery destination groups, a delivery route that travels to all of the delivery destinations in the delivery destination group without omission; program.

Images