Route selection device and route selection method

Abstract

A route selection device according to one aspect of the present invention comprises: an acquisition unit that acquires section environment data from each section between a start point and an end point; an event detection unit that uses the environment data to estimate, for each section, the degree of an adverse effect on a user's preferences; a cost setting unit that sets a cost for each section using the estimated degree of adverse effect on the user's preferences for each section; and a route estimation unit that selects, from a plurality of route candidates, the shortest route based on a section corresponding to the preferences of the user, on the basis of the set cost for each section.

Description

Route selection device and route selection method

[0001] The present invention relates to a route selection device and a route selection method.

[0002] Techniques for selecting and proposing routes have been developed. For example, Non-Patent Document 1 discloses a technique for recommending routes that tend to be similar to the route previously traveled, using two indices: the length of the road and the number of people passing by.

[0003] Junya Yoshida et al., "Route recommendation method reflecting mood during travel," 2nd Forum on Data Engineering and Information Management, A10-4, 2010

[0004] However, the technology described in Non-Patent Document 1 has a problem in that it is not possible to avoid upsetting a user due to an event that is undesirable for the individual user when presenting a route. In view of the above circumstances, an object of the present invention is to provide a technology that can avoid upsetting a user when presenting a route.

[0005] One aspect of the present invention is a route selection device that includes an acquisition unit that acquires environmental data for each section between a start point and an end point, an event detection unit that uses the environmental data to estimate the degree of adverse impact on user preferences for each section, a cost setting unit that sets a cost for each section using the estimated degree of adverse impact on user preferences for each section, and a route estimation unit that selects the shortest route based on sections that suit the user's preferences from a plurality of route candidates based on the set cost for each section.

[0006] One aspect of the present invention is a route selection method in which a route selection device acquires environmental data for each section between a start point and an end point, uses the environmental data to estimate the degree of adverse impact on user preferences for each section, sets a cost for each section using the estimated degree of adverse impact on user preferences for each section, and selects from a plurality of route candidates the shortest route based on sections that suit the user's preferences, based on the set cost for each section.

[0007] According to the present invention, it is possible to avoid upsetting the user when presenting a route.

[0008] FIG. 1 is a diagram illustrating an example of a route and an example of an event in route selection. FIG. 2 is a diagram illustrating an example of a configuration of a route selection device communication in the first embodiment. FIG. 3 is a diagram illustrating an example of an input and output of a model when used in the first embodiment. FIG. 4 is a diagram illustrating an example of an image captured by an environmental sensor. FIG. 5 is a flowchart illustrating an example of a processing procedure of a route selection system in the first embodiment. FIG. 6 is a diagram illustrating an example of a configuration of a model in the second embodiment. FIG. 7 is a flowchart illustrating an example of a processing procedure of model learning in the second embodiment.

[0009] An embodiment of the present invention will be described in detail with reference to the drawings. FIG. 1 is a diagram showing an example of a route and an example of an event in route selection. In FIG. 1, point ps is the starting point, and point pe is the ending point. The starting point ps is, for example, the current location or departure point. The ending point pe is, for example, the destination. Points p1 to p8 are waypoints or branching points. As shown in FIG. 1, there are multiple routes from the starting point ps to the ending point pe. Furthermore, each arrow ya represents a section (route segment) between the points.

[0010] In this embodiment, photographing devices (g1 to g3) that are environmental sensors are installed near any of the waypoints, branch points, and sections.

[0011] For example, in the example of FIG. 1 , a first image capturing device g1 installed near a first section between points p1 and p3 captures events in the first section. A second image capturing device g2 installed near a second section between points p5 and p7 captures events in the second section. A third image capturing device g3 installed near the first section between points p4 and p6 captures events in the third section. An event is a phenomenon that may affect a user's biometric information (e.g., vital signs) when the user passes through that section, such as a fight between people, littering, or darkness due to a malfunction of lighting equipment. In this embodiment, vital signs include, for example, pulse (heart rate), respiration, blood pressure, and body temperature.

[0012] A user may have preferences for routes, such as not liking crowds or wanting to avoid dark roads. However, a typical route selection device estimates and presents the shortest route, which does not necessarily match the user's preferences. For this reason, in conventional technology, events occurring in a section may upset the user's mood. For this reason, in an embodiment, events that are undesirable for each user are detected in real time or learned, and route selection is performed taking these events into consideration, thereby selecting a route that does not upset the user's mood. Note that in the embodiment, the user's movement may be by walking or by moving vehicle.

[0013] <First embodiment> Fig. 2 is a diagram showing an example of the configuration of a route selection system according to this embodiment. As shown in Fig. 2, the route selection system 1 includes, for example, a route selection device 2, an environmental sensor 3, a DB 4, and a mobile terminal 5. The route selection device 2 is connected to the environmental sensor 3, the DD 4, and the mobile terminal 5 via a wired or wireless network NW.

[0014] The route selection device 2 includes, for example, an acquisition unit 21, a route segmentation unit 22, an event detection unit 23, a cost setting unit 25, a route estimation unit 26, a route update unit 27, and an output unit 28. The environmental sensors 3 include n (n is an integer equal to or greater than 1) first environmental sensors 31-1 to n-th environmental sensors 31-n. The portable terminal 5 includes, for example, a GPS device 51, a photographing unit 52, an operation unit 53, a display unit 54, a sound collection unit 55, a communication unit 56, and a sensor 57.

[0015] The environmental sensor 3 is a sensor that detects environmental information, such as a photographing device, a sound collecting device, etc. In the following example, an example in which the environmental sensor 3 is a photographing device will be described.

[0016] The first environmental sensor 31-1 to the nth environmental sensor 31-n may be installed near all sections (or waypoints, branching points), or may be installed near specific sections (or waypoints, branching points). The first environmental sensor 31-1 to the nth environmental sensor 31-n may be equipped with a sound collection device (microphone) for collecting acoustic signals in addition to an imaging device. The first environmental sensor 31-1 to the nth environmental sensor 31-n are capable of constantly outputting data and are constantly connectable to the route selection device 2. The first environmental sensor 31-1 to the nth environmental sensor 31-n associate a captured image with identification information for identifying the environmental sensor 3 that captured the image and the time of capture, and output the associated image to the route selection device 2. In the embodiment, at least one of the image data and acoustic signals captured or collected by the first environmental sensor 31-1 to the nth environmental sensor 31-n is referred to as environmental data.

[0017] The mobile terminal 5 is, for example, any one of a smartphone, a tablet terminal, a car navigation device, a dedicated device, and the like.

[0018] The GPS device 51 obtains the location information of the mobile terminal 5 based on information acquired from a GPS (Global Positioning System).

[0019] The photographing unit 52 is a camera. When the model 24 is being trained, the photographing unit 52 takes an image in response to a user operation and outputs the taken image to the route selecting device 2. The training of the model 24 will be described in a second embodiment.

[0020] The operation unit 53 is, for example, a touch panel sensor provided on the display unit 54, and detects the results of operations performed by the user.

[0021] The display unit 54 displays the route information received by the communication unit 56. The display unit 54 displays a screen for allowing the user to select or input preferences when learning the model 24, and images and the like for learning preferences.

[0022] The sound pickup unit 55 is a microphone that picks up the user's voice signal.

[0023] The communication unit 56 transmits the location information determined by the GPS device 51, operation information resulting from the user's operation detected by the operation unit 53, and the audio signal collected by the sound collection unit 55 to the route selection device 3. The communication unit 56 acquires the route information output by the route selection device 2.

[0024] The sensor 57 is a detection device that detects biological information. The sensor 57 may be provided in, for example, a smart watch that can communicate with the mobile terminal 5.

[0025] DB4 is a database that stores, for example, map information. The map information includes location information (latitude and longitude) of each waypoint and branch point, location information of each start point and end point of a section, etc. DB4 associates identification information of the environmental sensors 31-n with installation location information that indicates the location where each environmental sensor 31-n is installed and stores the information.

[0026] The acquisition unit 21 acquires image data, which is environmental data, from the first environmental sensor 31-1 to the n-th environmental sensor 31-n, for example, at predetermined time intervals. The acquisition unit 21 acquires route information from the DB 4. The acquisition unit 21 acquires location information and operation information (or audio signals) from the mobile terminal 5. When learning the model 24, the acquisition unit 21 acquires captured image data and biometric information from the mobile terminal 5.

[0027] The route segmentation unit 22 acquires position information, operation information (or audio signals), and route information from the acquisition unit 21. The route segmentation unit 22 sets the position information of the user's current location and end point (destination) using the information acquired from the acquisition unit 21. The route segmentation unit 22 receives the user's current location and end point as input, divides multiple routes connecting the start point and end point into certain sections, and outputs the section information. The section information includes information indicating the length of each section. The route segmentation unit 22 may acquire the position information of the user's current location by any means. For example, the route segmentation unit 22 may use information from the GPS device 51 included in the user's mobile terminal 5. The division of the route may be determined in any manner. The division may be determined to be a certain distance, or may be determined based on corners or buildings. The route segmentation unit 22 may acquire the position information of the user's end point by any means. The route division unit 22 may, for example, acquire location information of the end point based on the result of an operation performed on the operation unit 53 of the user's mobile terminal 5, or may acquire an audio signal picked up by the sound pickup unit 55 of the mobile terminal and perform voice recognition on the acquired audio signal to acquire location information of the end point.

[0028] The event detection unit 23 includes a model 24. During learning, the event detection unit 23 trains the model 24 in advance using, for example, image data and information indicating the degree of adverse impact that the image information has on, for example, the user's vital signs as training data. During use, as shown in FIG. 3 , the event detection unit 23 inputs image data acquired from an environmental sensor 3 installed near the section identified by the route division unit 22 into the trained model 24, and outputs an estimation result of the degree of adverse impact that each section has on, for example, the user's vital signs. FIG. 3 is a diagram showing an example of the input and output of the model during use in this embodiment. The model 24 is, for example, configured as a neural network.

[0029] The cost setting unit 25 sets a cost for each section based on the length of the section and the estimated result of the degree of adverse effect that the section will have on the user's vital signs output by the event detection unit 23. The length of the section may be stored in DB4, or may be calculated by the cost setting unit 25 using the position information of the start point and end point of the section. There are no particular limitations on the method for setting the cost. For example, the cost setting method may be to normalize and add the distance and the estimated degree of adverse effect, or to consider the distance to be longer by the estimated degree of adverse effect.

[0030] The route estimation unit 26 selects the shortest route based on the cost of each section output by the cost setting unit 25. Any method may be used to determine the shortest route. For example, the method of determining the shortest route may be to derive the shortest route using dynamic programming. In this embodiment, instead of selecting a general route with the shortest distance, a route that avoids sections that the user does not like and passes through other sections is selected, which is the shortest route.

[0031] For example, after a certain time has passed since the route estimation unit 26 output is received, the route update unit 27 repeats the process performed by the route estimation unit 26 from the route segmentation unit 22 to reselect the shortest route. The route update unit 27 regards the point where the user is located at the time of the update as the starting point of the input to the route segmentation unit. Note that the certain time after the route estimation unit 26 output is received may be, for example, after a certain distance (number of sections) has been traveled, or may be set as a time period. The reason for updating in this manner is that an event may occur after the user has moved, even if it did not occur when the user started moving.

[0032] The output unit 28 outputs to the mobile terminal 5 information indicating the shortest route determined by the route estimation unit 26 during the first processing, and outputs to the mobile terminal 5 information indicating the shortest route updated by the route update unit 27 after the update.

[0033] 2 is merely an example and is not limiting. For example, the route selection device 2 may include the DB 4. Alternatively, the functions of the route selection device 2 may be executed on a cloud. Alternatively, the mobile terminal 5 may include all or part of the functions of the route selection device 2.

[0034] The route selection device 2 is configured using a processor such as a CPU (Central Processing Unit) and a memory. When the processor executes a program, the route selection device 2 functions as a route division unit 22, an event detection unit 23, a cost setting unit 25, a route estimation unit 26, and a route update unit 27. All or part of the functions of the route selection device 2 may be implemented using hardware such as an ASIC (Application Specific Integrated Circuit), a PLD (Programmable Logic Device), or an FPGA (Field Programmable Gate Array). The program may be recorded on a computer-readable recording medium. Examples of computer-readable recording media include portable media such as a flexible disk, a magneto-optical disk, a ROM, a CD-ROM, and a semiconductor storage device (e.g., a solid-state drive (SSD)), as well as storage devices such as a hard disk or semiconductor storage device built into a computer system. The program may be transmitted via a telecommunications line.

[0035] 4 is a diagram showing examples of images captured by the environmental sensor. Image g101 is an example of an image captured in an area where litter is scattered. Image g102 is an example of an image captured in an area where litter is not scattered. Image g103 is an example of an image captured in an area where streetlights are out of order and it is dark. Image g104 is an example of an image captured in an area where it is bright.

[0036] The timing at which the route selection device 2 acquires images from the environmental sensor 3 is when the user requests route information and each time the process is performed multiple times until the user reaches the destination, as described above. The route selection device 2 may be configured not to acquire images of sections that the user has already passed through.

[0037] 5 is a flowchart of an example of a processing procedure of the route selection system of this embodiment. (Step S1) The acquisition unit 21 acquires location information and operation information (or voice signals) from the mobile terminal 5. The route division unit 22 sets the location information of the user's current location and end point (destination) using the information acquired from the acquisition unit 21.

[0038] (Step S2) The route division unit 22 receives the user's current location and end point as input, divides a plurality of routes connecting the start point and end point into certain sections, and generates section information.

[0039] (Step S3) The acquisition unit 21 acquires image data for each section from the environmental sensor 3 installed near the section.

[0040] (Step S4) The event detection unit 23 inputs image data acquired from the environmental sensor 3 installed near the section into the trained model 24, thereby estimating the degree of adverse impact that each section has on the user, for example, on their vital signs.

[0041] (Step S5) The cost setting unit 25 sets a cost for each section based on the length of the section and the estimated result output by the event detection unit 23 of the degree of adverse effect that the section will have on the user, for example, on vital signs.

[0042] (Step S6) If the processing is the first time, the route estimation unit 26 selects the shortest route based on the cost of each section output by the cost setting unit 25. If the processing is an update, the route update unit 27, after a certain period of time has elapsed since receiving the output of the route estimation unit 26, repeats the processing performed by the route estimation unit 26 from the route segmentation unit 22, and reselects the shortest route.

[0043] (Step S7) If the processing is the first time, the output unit 28 outputs information indicating the shortest route determined by the route estimation unit 26 to the mobile terminal 5. If the processing is an update, the output unit 28 outputs information indicating the shortest route updated by the route update unit 27 to the mobile terminal 5.

[0044] (Step S8) The route update unit 27 determines whether the user has arrived at the destination (end point) based on, for example, whether the current location is within a predetermined range of the destination. If the route update unit 27 determines that the user has arrived at the destination (end point) (Step S8; YES), the process ends. If the route update unit 27 determines that the user has not arrived at the destination (end point) (Step S8; NO), the process proceeds to Step S9.

[0045] (Step S9) The acquisition unit 21 acquires location information from the portable terminal 5. The route division unit 22 sets location information of the user's current location using the information acquired from the acquisition unit 21. After the process, the route division unit 22 proceeds to the process of step S2.

[0046] The processing procedure described with reference to Fig. 5 is an example, and is not limited to this. Some processes may be performed in parallel.

[0047] As described above, in this embodiment, the shortest route that avoids sections that the user does not like and passes through other sections is selected from multiple route candidates. As a result, according to this embodiment, it is possible to avoid upsetting the user when presenting routes.

[0048] In addition, in this embodiment, the route selection is updated when the user moves, so that the shortest route can be reselected and presented in response to an event that occurs during the user's movement.

[0049] The estimation result by the model 24 may be configured to output not only the negative impact but also the degree of favorability. In this case, the shortest route may be selected by avoiding sections that the user does not like and taking into account sections that the user likes (for example, by increasing the weighting value).

[0050] Second Embodiment In this embodiment, an example of learning a model 24 will be described. Fig. 6 is a diagram showing an example of the configuration of a model in this embodiment. The model 24 includes, for example, an estimation unit 241 and a parameter update unit 242. As shown in Fig. 6, during learning, image data for learning is input to the model 24, and the model 24 outputs the degree of adverse effect.

[0051] The image data for learning is, for example, image data that represents an event that the user encounters in daily life, captured by the image capture unit 52 of the mobile terminal 5, for example.

[0052] The training data is information obtained by acquiring vital signs when the image data is captured by the sensor 57 and normalizing the change in the acquired vital signs over several levels. The sensor 57 is a device capable of acquiring biometric information, and may be, for example, a smartwatch. The information obtained by normalizing the change in vital signs over several levels is one example of environmental preference information, and may be information selected by the user from several levels of evaluation values ​​displayed on the display unit 54 of the mobile terminal 5.

[0053] The portable terminal 5 associates the image data with the teacher data to create a set of learning data. The portable terminal 5 may transmit the created learning data to the route selection device 2 each time it is created. The learning data is assigned identification information that can identify the user. The route selection device 2 may store the acquired learning data for each user on its own device, or may store it in DB 4.

[0054] The estimation unit 241 is configured, for example, by a neural network. The event detection unit 24 sets initial values ​​in advance for each parameter of the neural network. The estimation unit 241 receives learning image data included in the learning data and outputs an estimated value of the degree.

[0055] The parameter update unit 242 updates each parameter of the neural network of the estimation unit 241 based on the comparison result between the estimated value of the degree output by the estimation unit 241 and the teacher data corresponding to the learning image data included in the learning data. Note that the parameter update may use, for example, a well-known error propagation learning method.

[0056] 7 is a flowchart of an example of a process procedure for learning a model according to this embodiment. (Step S11) The acquisition unit 21 acquires learning data.

[0057] (Step S12) The event detection unit 24 sets initial values ​​in advance for each parameter of the neural network.

[0058] (Step S13) The estimation unit 241 receives the learning image data included in the learning data as input and outputs an estimated value of the degree.

[0059] (Step S14) The parameter update unit 242 compares the estimated value of the degree output by the estimation unit 241 with the teacher data corresponding to the learning image data included in the learning data.

[0060] (Step S15) The parameter update unit 242 updates each parameter of the neural network of the estimation unit 241 based on the comparison result.

[0061] (Step S16) The parameter update unit 242 determines whether learning has ended. If the parameter update unit 242 determines that learning has ended (step S16; YES), it ends the learning process. If the parameter update unit 242 determines that learning has not ended (step S16; NO), it returns to the process of step S13. The parameter update unit 242 repeatedly executes the parameter update process until, for example, the comparison result satisfies a predetermined criterion. Note that the predetermined criterion is, for example, a state in which the number of repetitions has reached a predetermined number, or a state in which the comparison result is below a predetermined threshold, etc.

[0062] As described above, in this embodiment, the model 24 is configured to learn in advance the image data captured by the user and the degree of change in the user's vital signs. This allows the model 24 to learn the user's preferences for the route, thereby reflecting the user's preferences for events occurring along the route.

[0063] The route selection device 2 may determine the road congestion state, such as traffic volume, from the captured image, or may acquire information about traffic volume from a server that provides information about traffic volume, for example.

[0064] In the above example, the model 24 is learned in advance, but this is not limiting. The user may, for example, operate the operation unit 53 of the mobile terminal 5 to input the degree of preference for the currently traversed section. In such a case, the route selection unit 2 may update the model 24 using image data based on the current location and the degree of preference for the currently traversed section.

[0065] As described above, in the first and second embodiments, the degree of deterioration of a person's vital signs, for example, when facing each event is estimated based on the user's individual characteristics. Furthermore, in the first and second embodiments, an optimal route is selected that reflects the user's preferences by acquiring nearby video in real time. Note that these preferences are not generalized averages but are closer to individual tastes. As a result, in the first and second embodiments, not only road information such as traffic volume but also the user's preferences regarding events occurring along the route can be reflected. Furthermore, in the first and second embodiments, updating the selected route can realize more optimal route recommendations.

[0066] (Modification) In the second embodiment, the image data for learning may be an acoustic signal. In this case, the environmental sensor 3 may be a sound collection device, and the model 24 may be learned using training data on the acoustic signal and the degree of sound dislike. As a result, according to this modification, it is possible to select a route that avoids sections where events involving sounds that the user dislikes (such as noises or animal cries) occur.

[0067] The route selection system 1 may also include a sensor that detects information about odors. In this case, the environmental sensor 3 is a sensor that detects odors, and in the second embodiment, the image data for learning may be the degree to which the user dislikes odors. In this case, the model 24 may be trained using training data about odors and the degree to which the user dislikes odors. As a result, this modification allows the user to select a route that avoids sections where events involving odors that the user dislikes (such as food odors, odors from factories, garbage odors, etc.) occur.

[0068] The environmental data (image data, sound signals, and information related to smells) in the above-described modified example is merely an example, and any data that may potentially upset the user's mood during route selection may be used.

[0069] Although an embodiment of the present invention has been described above in detail with reference to the drawings, the specific configuration is not limited to this embodiment, and includes designs within the scope of the gist of the present invention.

[0070] The present invention can be applied to devices and systems that select and present routes for people or moving objects.

[0071] 1...Route selection system, 2...Route selection device, 3...Environmental sensor, 4...DB, 5...Mobile terminal, NW...Network, 21...Acquisition unit, 22...Route classification unit, 23...Event detection unit, 24...Model, 25...Cost setting unit, 26...Route estimation unit, 27...Route update unit, 28...Output unit, 31-1...First environmental sensor, 31-n...nth environmental sensor, 51...GPS device, 52...Photographing unit, 53...Operation unit, 54...Display unit, 55...Sound collection unit, 56...Communication unit, 57...Sensor, 241...Estimation unit, 242...Parameter update unit

Claims

1. A route selection device comprising: an acquisition unit that acquires environmental data for each section between a start point and an end point; an event detection unit that uses the environmental data to estimate the degree of adverse impact on user preferences for each section; a cost setting unit that sets a cost for each section using the estimated degree of adverse impact on the user preferences for each section; and a route estimation unit that selects the shortest route based on the section in accordance with the user's preferences from a plurality of route candidates based on the set cost for each section.

2. The route selection device according to claim 1, wherein the shortest route based on the sections according to the user's preferences is a route that passes through other sections that avoid sections that are not preferred by the user.

3. The route selection device of claim 1, wherein the event detection unit is provided with a model, inputs the environmental data into the model and outputs an estimated degree value, compares the estimated degree value with correct degree data corresponding to the environmental data, and learns the model by repeatedly updating each parameter of the model a predetermined number of times based on the results of the comparison.

4. A route selection method in which a route selection device acquires environmental data for each section between a starting point and an end point, estimates the degree of adverse impact on user preferences for each section using the environmental data, sets a cost for each section using the estimated degree of adverse impact on user preferences for each section, and selects from a plurality of route candidates a route that is the shortest based on the section that meets the user's preferences based on the set cost for each section.

Images